Install valgrind for Linux github runners

Allow user code to access specific pvalue values, especially when
multiple ptypes are in use.

.github/workflows/run-tests.yml

@@ -0,0 +1,38 @@
+name: Run Propane Tests
+
+on:
+  push:
+    branches:
+    - master
+  pull_request:
+
+jobs:
+  test:
+    runs-on: ${{ matrix.os }}
+    strategy:
+      matrix:
+        os: [ubuntu-latest, macos-latest]
+        ruby-version: ['3.4']
+
+    steps:
+    - name: Install dependencies (Linux)
+      if: runner.os == 'Linux'
+      run: sudo apt-get update && sudo apt-get install -y gcc gdc ldc valgrind
+
+    - name: Install dependencies (macOS)
+      if: runner.os == 'macOS'
+      run: brew install gcc ldc
+
+    - name: Checkout repository
+      uses: actions/checkout@v4
+
+    - name: Set up Ruby
+      uses: ruby/setup-ruby@v1
+      with:
+        ruby-version: ${{ matrix.ruby-version }}
+
+    - name: Install dependencies
+      run: bundle install
+
+    - name: Run tests
+      run: rake all

.rspec

@@ -1,3 +1,2 @@
---format documentation
 --color
 --require spec_helper

CHANGELOG.md

@@ -1,3 +1,122 @@
+## v4.7.0
+
+### New Features
+
+- Support parser rule user code blocks in tree generation mode.
+
+### Fixes
+
+- propane.vim: do not highlight rule components as propane keywords
+
+## v4.6.0
+
+### New Features
+
+- Add lexer user code API to access matched input text positions
+- Track rule component text positions and add parser user code API to access
+
+### Fixes
+
+- Fixed a few user guide and source comments related to text input positions
+
+## v4.5.0
+
+### New Features
+
+- Add `noline` grammar statement to skip emitting `#line` directives
+- Attempt to autodetect target language (D/C++) in extra/vim/syntax/propane.vim
+
+### Fixes
+
+- Fix #line reset directives
+- Update keyword list in extra/vim/syntax/propane.vim
+- Fix propane.vim keyword detection
+
+## v4.4.0
+
+### New Features
+
+- Add p_value_get() / p_value_get_XXX() accessors
+
+## v4.3.0
+
+### New Features
+
+- Use #line for user code blocks to report input grammar position for errors.
+
+## v4.2.0
+
+### New Features
+
+- Add support for a custom lex function.
+
+## v4.1.0
+
+### New Features
+
+- Add `p_context_delete()` and `p_tree_delete()` for D targets.
+
+## v4.0.0
+
+### New Features
+
+- Add `context_user_fields` statement to allow custom context user fields.
+- Add `token_user_fields` statement to allow custom token user fields.
+- Add `on_token_node` statement to allow custom code when constructing token nodes.
+- Add `free_token_node` statement to allow custom code when freeing token nodes.
+- Add `p_context_delete()`.
+- Allow `drop` patterns to execute lexer user code blocks.
+
+### Breaking Changes
+
+- Replace `p_context_init()` with `p_context_new()` and `p_context_delete()`.
+- Renamed `p_free_tree()` to `p_tree_delete()`.
+- The `free_token_node` statement now takes a user code block instead of a
+  function name parameter.
+
+## v3.0.0
+
+### New Features
+
+- Add support for multiple starting rules (#38)
+- Add `p_free_tree()` functions to reclaim generated tree memory
+- Add `free_token_node` grammar statement to reclaim user-allocated memory stored in a Token tree node `pvalue` field
+- Add valgrind memory leak tests to unit tests
+- Fix build issues for C++ to officially support C++ target output
+
+### Improvements
+
+- Document `p_lex()` and `p_token_info_t` in user guide (#37)
+
+### Breaking Changes
+
+- Rename AST generation mode to tree generation mode (see [UPGRADING.md](UPGRADING.md))
+
+## v2.3.0
+
+### New Features
+
+- Add \D, \S, \w, \W special character classes
+
+### Improvements
+
+- Include line numbers for pattern errors
+- Improve performance in a few places
+- Parallelize parser table generation on Linux hosts
+- Add github workflow to run unit tests
+
+### Fixes
+
+- Fix a couple clang warnings for C backend
+- Fix C backend not fully initializing pvalues when multiple ptypes are used with different sizes.
+- Fix some user guide examples
+
+## v2.2.1
+
+### Fixes
+
+- Fix GC issue for D backend when AST is enabled (#36)
+
 ## v2.2.0
 
 ### Improvements

Gemfile.lock

@@ -1,38 +1,42 @@
 GEM
   remote: https://rubygems.org/
   specs:
-    base64 (0.2.0)
-    date (3.4.1)
-    diff-lcs (1.5.1)
+    base64 (0.3.0)
+    date (3.5.1)
+    diff-lcs (1.6.2)
     docile (1.4.1)
-    psych (5.2.2)
+    erb (6.0.4)
+    psych (5.4.0)
       date
       stringio
-    rake (13.2.1)
-    rdoc (6.10.0)
+    rake (13.4.2)
+    rdoc (7.2.0)
+      erb
       psych (>= 4.0.0)
-    redcarpet (3.6.0)
-    rspec (3.13.0)
+      tsort
+    redcarpet (3.6.1)
+    rspec (3.13.2)
       rspec-core (~> 3.13.0)
       rspec-expectations (~> 3.13.0)
       rspec-mocks (~> 3.13.0)
-    rspec-core (3.13.2)
+    rspec-core (3.13.6)
       rspec-support (~> 3.13.0)
-    rspec-expectations (3.13.3)
+    rspec-expectations (3.13.5)
       diff-lcs (>= 1.2.0, < 2.0)
       rspec-support (~> 3.13.0)
-    rspec-mocks (3.13.2)
+    rspec-mocks (3.13.8)
       diff-lcs (>= 1.2.0, < 2.0)
       rspec-support (~> 3.13.0)
-    rspec-support (3.13.2)
+    rspec-support (3.13.7)
     simplecov (0.22.0)
       docile (~> 1.1)
       simplecov-html (~> 0.11)
       simplecov_json_formatter (~> 0.1)
-    simplecov-html (0.13.1)
+    simplecov-html (0.13.2)
     simplecov_json_formatter (0.1.4)
-    stringio (3.1.2)
+    stringio (3.2.0)
     syntax (1.2.2)
+    tsort (0.2.0)
 
 PLATFORMS
   ruby
@@ -47,4 +51,4 @@ DEPENDENCIES
   syntax
 
 BUNDLED WITH
-   2.3.7
+  4.0.14

LICENSE.txt

@@ -1,6 +1,6 @@
 The MIT License (MIT)
 
-Copyright (c) 2010-2024 Josh Holtrop
+Copyright (c) 2010-2026 Josh Holtrop
 
 Permission is hereby granted, free of charge, to any person obtaining a copy
 of this software and associated documentation files (the "Software"), to deal

README.md

@@ -6,8 +6,10 @@ Propane is a LALR Parser Generator (LPG) which:
   * generates a built-in lexer to tokenize input
   * supports UTF-8 lexer inputs
   * generates a table-driven shift/reduce parser to parse input in linear time
-  * targets C or D language outputs
-  * optionally supports automatic full AST generation
+  * targets C, C++, or D language outputs
+  * optionally supports automatic full parse tree generation
+  * supports starting parsing from multiple start rules
+  * tracks input text start and end positions for all matched tokens/rules
   * is MIT-licensed
   * is distributable as a standalone Ruby script
 

Rakefile.rb

@@ -1,5 +1,7 @@
 require "rake/clean"
 require "rspec/core/rake_task"
+require "simplecov"
+require "stringio"
 
 CLEAN.include %w[spec/run gen .yardoc yard coverage dist]
 
@@ -12,6 +14,18 @@ RSpec::Core::RakeTask.new(:spec, :example_pattern) do |task, args|
     task.rspec_opts = %W[-e "#{args.example_pattern}" -f documentation]
   end
 end
+task :spec do |task, args|
+  unless ENV["dist_specs"]
+    original_stdout = $stdout
+    sio = StringIO.new
+    $stdout = sio
+    SimpleCov.collate Dir["coverage/.resultset.json"]
+    $stdout = original_stdout
+    sio.string.lines.each do |line|
+      $stdout.write(line) unless line =~ /Coverage report generated for/
+    end
+  end
+end
 
 # dspec task is useful to test the distributable release script, but is not
 # useful for coverage information.

UPGRADING.md

@@ -0,0 +1,21 @@
+## v4.0.0
+
+### API Changes
+
+- Replace any calls to `p_context_init()` with `p_context_new()`.
+- Replace any references to the address of a statically allocated context
+  structure with the pointer returned from `p_context_init()` (e.g. `&context`
+  -> `context`).
+- Add a call to `p_context_delete()` (for C or C++) after lexing/parsing to
+  reclaim context memory.
+- Rename `p_free_tree()` calls to `p_tree_delete()`.
+- Change `free_token_node` statement calls from taking a function name argument
+  to taking a user code block.
+
+## v3.0.0
+
+### Grammar Changes
+
+- Rename `ast;` statement to `tree;`.
+- Rename `ast_prefix;` statement to `tree_prefix;`.
+- Rename `ast_suffix;` statement to `tree_suffix;`.

assets/parser.c.erb

@@ -43,29 +43,52 @@ const char * <%= @grammar.prefix %>token_names[] = {
  *************************************************************************/
 
 /**
- * Initialize lexer/parser context structure.
+ * Allocate and initialize lexer/parser context structure.
+ *
+ * Deinitialize and deallocate with <%= @grammar.prefix %>context_delete().
  *
- * @param[out] context
- *   Lexer/parser context structure.
  * @param input
  *   Text input.
  * @param input_length
  *   Text input length.
+ *
+ * @return Context structure for lexer/parser.
  */
-void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, uint8_t const * input, size_t input_length)
+<%= @grammar.prefix %>context_t * <%= @grammar.prefix %>context_new(uint8_t const * input, size_t input_length)
 {
-    /* New default-initialized context structure. */
-    <%= @grammar.prefix %>context_t newcontext = {0};
+<% if @cpp %>
+    <%= @grammar.prefix %>context_t * context = new <%= @grammar.prefix %>context_t();
+<% else %>
+    <%= @grammar.prefix %>context_t * context = (<%= @grammar.prefix %>context_t *)calloc(1, sizeof(<%= @grammar.prefix %>context_t));
+<% end %>
 
     /* Lexer initialization. */
-    newcontext.input = input;
-    newcontext.input_length = input_length;
-    newcontext.text_position.row = 1u;
-    newcontext.text_position.col = 1u;
-    newcontext.mode = <%= @lexer.mode_id("default") %>;
+    context->input = input;
+    context->input_length = input_length;
+    context->text_position.row = 1u;
+    context->text_position.col = 1u;
+    context->mode = <%= @lexer.mode_id("default") %>;
 
-    /* Copy to the user's context structure. */
-    *context = newcontext;
+    return context;
+}
+
+/**
+ * Deinitialize and deallocate lexer/parser context structure.
+ *
+ * For C++, destructors will be called for any context user fields. However, if
+ * pointers are used to store allocated resources, the user should free them
+ * before calling this function.
+ *
+ * @param context
+ *   Lexer/parser context structure allocated with <%= @grammar.prefix %>context_new().
+ */
+void <%= @grammar.prefix %>context_delete(<%= @grammar.prefix %>context_t * context)
+{
+<% if @cpp %>
+    delete context;
+<% else %>
+    free(context);
+<% end %>
 }
 
 /**************************************************************************
@@ -344,8 +367,10 @@ static lexer_state_id_t check_lexer_transition(uint32_t current_state, uint32_t
 static size_t find_longest_match(<%= @grammar.prefix %>context_t * context,
     lexer_match_info_t * out_match_info, size_t * out_unexpected_input_length)
 {
-    lexer_match_info_t longest_match = {0};
-    lexer_match_info_t attempt_match = {0};
+    lexer_match_info_t longest_match;
+    memset(&longest_match, 0, sizeof(longest_match));
+    lexer_match_info_t attempt_match;
+    memset(&attempt_match, 0, sizeof(attempt_match));
     *out_match_info = longest_match;
     uint32_t current_state = lexer_mode_table[context->mode].state_table_offset;
     for (;;)
@@ -359,6 +384,7 @@ static size_t find_longest_match(<%= @grammar.prefix %>context_t * context,
         switch (result)
         {
         case P_SUCCESS:
+            {
             lexer_state_id_t transition_state = check_lexer_transition(current_state, code_point);
             if (transition_state != INVALID_LEXER_STATE_ID)
             {
@@ -390,6 +416,7 @@ static size_t find_longest_match(<%= @grammar.prefix %>context_t * context,
                 *out_unexpected_input_length = attempt_match.length + code_point_length;
                 return P_UNEXPECTED_INPUT;
             }
+            }
             break;
 
         case P_EOF:
@@ -447,7 +474,8 @@ static size_t find_longest_match(<%= @grammar.prefix %>context_t * context,
  */
 static size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%= @grammar.prefix %>token_info_t * out_token_info)
 {
-    <%= @grammar.prefix %>token_info_t token_info = {0};
+    <%= @grammar.prefix %>token_info_t token_info;
+    memset(&token_info, 0, sizeof(token_info));
     token_info.position = context->text_position;
     token_info.token = INVALID_TOKEN_ID;
     lexer_match_info_t match_info;
@@ -456,7 +484,24 @@ static size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%= @
     switch (result)
     {
     case P_SUCCESS:
+        {
         <%= @grammar.prefix %>token_t token_to_accept = match_info.accepting_state->token;
+        /* Calculate the token length and start/end positions before invoking
+         * the lexer user code so that the user code can access them. The
+         * context input text position tracking is not updated until after the
+         * user code has run so that it is left unchanged if the user code
+         * requests to terminate the lexer. */
+        token_info.length = match_info.length;
+        if (match_info.end_delta_position.row != 0u)
+        {
+            token_info.end_position.row = token_info.position.row + match_info.end_delta_position.row;
+            token_info.end_position.col = match_info.end_delta_position.col;
+        }
+        else
+        {
+            token_info.end_position.row = token_info.position.row;
+            token_info.end_position.col = token_info.position.col + match_info.end_delta_position.col;
+        }
         if (match_info.accepting_state->code_id != INVALID_USER_CODE_ID)
         {
             uint8_t const * match = &context->input[context->input_index];
@@ -495,18 +540,8 @@ static size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%= @
             return P_DROP;
         }
         token_info.token = token_to_accept;
-        token_info.length = match_info.length;
-        if (match_info.end_delta_position.row != 0u)
-        {
-            token_info.end_position.row = token_info.position.row + match_info.end_delta_position.row;
-            token_info.end_position.col = match_info.end_delta_position.col;
-        }
-        else
-        {
-            token_info.end_position.row = token_info.position.row;
-            token_info.end_position.col = token_info.position.col + match_info.end_delta_position.col;
-        }
         *out_token_info = token_info;
+        }
         return P_SUCCESS;
 
     case P_EOF:
@@ -630,7 +665,7 @@ typedef struct
      * reduce action.
      */
     parser_state_id_t n_states;
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 
     /**
      * Map of rule components to rule set child fields.
@@ -638,7 +673,7 @@ typedef struct
     uint16_t const * rule_set_node_field_index_map;
 
     /**
-     * Number of rule set AST node fields.
+     * Number of rule set tree node fields.
      */
     uint16_t rule_set_node_field_array_size;
 
@@ -677,22 +712,28 @@ typedef struct
     /** Parser state ID. */
     size_t state_id;
 
+<% if @grammar.tree %>
+    /** tree node. */
+    void * tree_node;
+<% else %>
+    <%= @grammar.prefix %>position_t position;
+    <%= @grammar.prefix %>position_t end_position;
     /** Parser value from this state. */
     <%= @grammar.prefix %>value_t pvalue;
-
-<% if @grammar.ast %>
-    /** AST node. */
-    void * ast_node;
 <% end %>
 } state_value_t;
 
-/** Common AST node structure. */
-typedef struct
+<% if @grammar.tree %>
+/** Common tree node structure. */
+typedef struct TreeNode_s
 {
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
-    void * fields[];
-} ASTNode;
+    uint16_t n_fields;
+    uint8_t is_token;
+    struct TreeNode_s * fields[];
+} TreeNode;
+<% end %>
 
 /** Parser shift table. */
 static const shift_t parser_shift_table[] = {
@@ -701,7 +742,7 @@ static const shift_t parser_shift_table[] = {
 <%   end %>
 };
 
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 <%   @grammar.rules.each do |rule| %>
 <%     unless rule.flat_rule_set_node_field_index_map? %>
 const uint16_t r_<%= rule.name.gsub("$", "_") %><%= rule.id %>_node_field_index_map[<%= rule.rule_set_node_field_index_map.size %>] = {<%= rule.rule_set_node_field_index_map.map {|v| v.to_s}.join(", ") %>};
@@ -712,17 +753,22 @@ const uint16_t r_<%= rule.name.gsub("$", "_") %><%= rule.id %>_node_field_index_
 /** Parser reduce table. */
 static const reduce_t parser_reduce_table[] = {
 <%   @parser.reduce_table.each do |reduce| %>
-    {<%= reduce[:token_id] %>u, <%= reduce[:rule_id] %>u, <%= reduce[:rule_set_id] %>u, <%= reduce[:n_states] %>u
-<%     if @grammar.ast %>
+    {
+        <%= reduce[:token_id] %>u, /* Token: <%= reduce[:token] ? reduce[:token].name : "(any)" %> */
+        <%= reduce[:rule_id] %>u, /* Rule ID */
+        <%= reduce[:rule_set_id] %>u, /* Rule set ID (<%= reduce[:rule].rule_set.name %>) */
+<%     if @grammar.tree %>
+        <%= reduce[:n_states] %>u, /* Number of states */
 <%       if reduce[:rule].flat_rule_set_node_field_index_map? %>
-             , NULL
+        NULL, /* No rule set node field index map (flat map) */
 <%       else %>
-             , &r_<%= reduce[:rule].name.gsub("$", "_") %><%= reduce[:rule].id %>_node_field_index_map[0]
+        &r_<%= reduce[:rule].name.gsub("$", "_") %><%= reduce[:rule].id %>_node_field_index_map[0], /* Rule set node field index map */
 <%       end %>
-             , <%= reduce[:rule].rule_set.ast_fields.size %>
-             , <%= reduce[:propagate_optional_target] %>
+        <%= reduce[:rule].rule_set.tree_fields.size %>, /* Number of tree fields */
+        <%= reduce[:propagate_optional_target] %>}, /* Propagate optional target? */
+<%     else %>
+        <%= reduce[:n_states] %>u},
 <%     end %>
-    },
 <%   end %>
 };
 
@@ -792,7 +838,7 @@ static void state_values_stack_push(state_values_stack_t * stack)
     if (current_length >= current_capacity)
     {
         size_t const new_capacity = current_capacity * 2u;
-        state_value_t * new_entries = malloc(new_capacity * sizeof(state_value_t));
+        state_value_t * new_entries = (state_value_t *)malloc(new_capacity * sizeof(state_value_t));
         memcpy(new_entries, stack->entries, current_length * sizeof(state_value_t));
         free(stack->entries);
         stack->capacity = new_capacity;
@@ -826,7 +872,60 @@ static void state_values_stack_free(state_values_stack_t * stack)
      free(stack->entries);
 }
 
-<% unless @grammar.ast %>
+<% unless @grammar.tree %>
+/**
+ * Get the rule position (start or end) for the currently matched rule.
+ */
+static <%= @grammar.prefix %>position_t get_rule_position(state_values_stack_t * statevalues, size_t i, size_t n_states, bool get_end)
+{
+    if (n_states > 0u)
+    {
+        if (i == 0u)
+        {
+            if (get_end)
+            {
+                int stack_index = -1;
+                for (size_t j = 0u; j < n_states; j++)
+                {
+                    state_value_t * sv = state_values_stack_index(statevalues, stack_index - (int)j);
+                    if (<%= @grammar.prefix %>position_valid(sv->end_position))
+                    {
+                        return sv->end_position;
+                    }
+                }
+            }
+            else
+            {
+                int stack_index = -(int)n_states;
+                for (size_t j = 0u; j < n_states; j++)
+                {
+                    state_value_t * sv = state_values_stack_index(statevalues, stack_index + (int)j);
+                    if (<%= @grammar.prefix %>position_valid(sv->position))
+                    {
+                        return sv->position;
+                    }
+                }
+            }
+        }
+        else
+        {
+            if (get_end)
+            {
+                return state_values_stack_index(statevalues, -1 - (int)n_states + (int)i)->end_position;
+            }
+            else
+            {
+                return state_values_stack_index(statevalues, -1 - (int)n_states + (int)i)->position;
+            }
+        }
+    }
+    <%= @grammar.prefix %>position_t empty_pos;
+    memset(&empty_pos, 0, sizeof(empty_pos));
+    return empty_pos;
+}
+<% end %>
+
+<% if !@grammar.tree || @grammar.parser_user_code_used? %>
 /**
  * Execute user code associated with a parser rule.
  *
@@ -837,7 +936,7 @@ static void state_values_stack_free(state_values_stack_t * stack)
  * @retval P_USER_TERMINATED
  *   User requested to terminate parsing.
  */
-static size_t parser_user_code(<%= @grammar.prefix %>value_t * _pvalue, uint32_t rule, state_values_stack_t * statevalues, uint32_t n_states, <%= @grammar.prefix %>context_t * context)
+static size_t parser_user_code(<%= @grammar.tree ? "void" : "#{@grammar.prefix}value_t" %> * _pvalue, uint32_t rule, state_values_stack_t * statevalues, uint32_t n_states, <%= @grammar.prefix %>context_t * context)
 {
     switch (rule)
     {
@@ -909,6 +1008,8 @@ static size_t check_reduce(size_t state_id, <%= @grammar.prefix %>token_t token)
  *
  * @param context
  *   Lexer/parser context structure.
+ * @start_state_id
+ *   ID of the state in which to start.
  *
  * @retval P_SUCCESS
  *   The parser successfully matched the input text. The parse result value
@@ -921,25 +1022,28 @@ static size_t check_reduce(size_t state_id, <%= @grammar.prefix %>token_t token)
  * @reval P_UNEXPECTED_INPUT
  *   Input text does not match any lexer pattern.
  */
-size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
+static size_t parse_from(<%= @grammar.prefix %>context_t * context, size_t start_state_id)
 {
     <%= @grammar.prefix %>token_info_t token_info;
     <%= @grammar.prefix %>token_t token = INVALID_TOKEN_ID;
     state_values_stack_t statevalues;
     size_t reduced_rule_set = INVALID_ID;
-<% if @grammar.ast %>
+<% if @grammar.tree %>
     void * reduced_parser_node;
 <% else %>
+    <%= @grammar.prefix %>position_t reduced_position;
+    <%= @grammar.prefix %>position_t reduced_end_position;
     <%= @grammar.prefix %>value_t reduced_parser_value;
 <% end %>
     state_values_stack_init(&statevalues);
     state_values_stack_push(&statevalues);
+    state_values_stack_index(&statevalues, -1)->state_id = start_state_id;
     size_t result;
     for (;;)
     {
         if (token == INVALID_TOKEN_ID)
         {
-            size_t lexer_result = <%= @grammar.prefix %>lex(context, &token_info);
+            size_t lexer_result = <%= lex_fn %>(context, &token_info);
             if (lexer_result != P_SUCCESS)
             {
                 result = lexer_result;
@@ -958,8 +1062,8 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
             if ((shift_state != INVALID_ID) && (token == TOKEN___EOF))
             {
                 /* Successful parse. */
-<% if @grammar.ast %>
-                context->parse_result = (<%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> *)state_values_stack_index(&statevalues, -1)->ast_node;
+<% if @grammar.tree %>
+                context->parse_result = state_values_stack_index(&statevalues, -1)->tree_node;
 <% else %>
                 context->parse_result = state_values_stack_index(&statevalues, -1)->pvalue;
 <% end %>
@@ -971,30 +1075,43 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
         {
             /* We have something to shift. */
             state_values_stack_push(&statevalues);
-            state_values_stack_index(&statevalues, -1)->state_id = shift_state;
+            state_value_t * new_state_info = state_values_stack_index(&statevalues, -1);
+            new_state_info->state_id = shift_state;
             if (reduced_rule_set == INVALID_ID)
             {
                 /* We shifted a token, mark it consumed. */
-<% if @grammar.ast %>
-                <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %> * token_ast_node = malloc(sizeof(<%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %>));
-                token_ast_node->position = token_info.position;
-                token_ast_node->end_position = token_info.end_position;
-                token_ast_node->token = token;
-                token_ast_node->pvalue = token_info.pvalue;
-                state_values_stack_index(&statevalues, -1)->ast_node = token_ast_node;
+<% if @grammar.tree %>
+<%   if @cpp %>
+                <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> * token_tree_node = new <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>();
 <%   else %>
-                state_values_stack_index(&statevalues, -1)->pvalue = token_info.pvalue;
+                <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> * token_tree_node = (<%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> *)malloc(sizeof(<%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>));
+<%   end %>
+                token_tree_node->position = token_info.position;
+                token_tree_node->end_position = token_info.end_position;
+                token_tree_node->n_fields = 0u;
+                token_tree_node->is_token = 1u;
+                token_tree_node->token = token;
+                token_tree_node->pvalue = token_info.pvalue;
+<%= expand_code(@grammar.on_token_node, false, nil, nil) %>
+                new_state_info->tree_node = token_tree_node;
+<% else %>
+                new_state_info->position = token_info.position;
+                new_state_info->end_position = token_info.end_position;
+                new_state_info->pvalue = token_info.pvalue;
 <% end %>
                 token = INVALID_TOKEN_ID;
             }
             else
             {
                 /* We shifted a RuleSet. */
-<% if @grammar.ast %>
-                state_values_stack_index(&statevalues, -1)->ast_node = reduced_parser_node;
+<% if @grammar.tree %>
+                new_state_info->tree_node = reduced_parser_node;
 <% else %>
-                state_values_stack_index(&statevalues, -1)->pvalue = reduced_parser_value;
-                <%= @grammar.prefix %>value_t new_parse_result = {0};
+                new_state_info->pvalue = reduced_parser_value;
+                new_state_info->position = reduced_position;
+                new_state_info->end_position = reduced_end_position;
+                <%= @grammar.prefix %>value_t new_parse_result;
+                memset(&new_parse_result, 0, sizeof(new_parse_result));
                 reduced_parser_value = new_parse_result;
 <% end %>
                 reduced_rule_set = INVALID_ID;
@@ -1006,39 +1123,38 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
         if (reduce_index != INVALID_ID)
         {
             /* We have something to reduce. */
-<% if @grammar.ast %>
+<% if @grammar.tree %>
             if (parser_reduce_table[reduce_index].propagate_optional_target)
             {
-                reduced_parser_node = state_values_stack_index(&statevalues, -1)->ast_node;
+                reduced_parser_node = state_values_stack_index(&statevalues, -1)->tree_node;
             }
             else if (parser_reduce_table[reduce_index].n_states > 0)
             {
                 size_t n_fields = parser_reduce_table[reduce_index].rule_set_node_field_array_size;
-                ASTNode * node = (ASTNode *)malloc(sizeof(ASTNode) + n_fields * sizeof(void *));
+                size_t bytes = sizeof(TreeNode) + n_fields * sizeof(void *);
+                TreeNode * node = (TreeNode *)malloc(bytes);
+                memset(node, 0, bytes);
                 node->position = INVALID_POSITION;
                 node->end_position = INVALID_POSITION;
-                for (size_t i = 0; i < n_fields; i++)
-                {
-                    node->fields[i] = NULL;
-                }
+                node->n_fields = n_fields;
                 if (parser_reduce_table[reduce_index].rule_set_node_field_index_map == NULL)
                 {
                     for (size_t i = 0; i < parser_reduce_table[reduce_index].n_states; i++)
                     {
-                        node->fields[i] = state_values_stack_index(&statevalues, -(int)parser_reduce_table[reduce_index].n_states + (int)i)->ast_node;
+                        node->fields[i] = (TreeNode *)state_values_stack_index(&statevalues, -(int)parser_reduce_table[reduce_index].n_states + (int)i)->tree_node;
                     }
                 }
                 else
                 {
                     for (size_t i = 0; i < parser_reduce_table[reduce_index].n_states; i++)
                     {
-                        node->fields[parser_reduce_table[reduce_index].rule_set_node_field_index_map[i]] = state_values_stack_index(&statevalues, -(int)parser_reduce_table[reduce_index].n_states + (int)i)->ast_node;
+                        node->fields[parser_reduce_table[reduce_index].rule_set_node_field_index_map[i]] = (TreeNode *)state_values_stack_index(&statevalues, -(int)parser_reduce_table[reduce_index].n_states + (int)i)->tree_node;
                     }
                 }
                 bool position_found = false;
                 for (size_t i = 0; i < n_fields; i++)
                 {
-                    ASTNode * child = (ASTNode *)node->fields[i];
+                    TreeNode * child = node->fields[i];
                     if ((child != NULL) && <%= @grammar.prefix %>position_valid(child->position))
                     {
                         if (!position_found)
@@ -1055,13 +1171,32 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
             {
                 reduced_parser_node = NULL;
             }
+<%   if @grammar.parser_user_code_used? %>
+            if (parser_user_code(reduced_parser_node, parser_reduce_table[reduce_index].rule, &statevalues, parser_reduce_table[reduce_index].n_states, context) == P_USER_TERMINATED)
+            {
+                state_values_stack_free(&statevalues);
+                return P_USER_TERMINATED;
+            }
+<%   end %>
 <% else %>
-            <%= @grammar.prefix %>value_t reduced_parser_value2 = {0};
+            <%= @grammar.prefix %>value_t reduced_parser_value2;
+            memset(&reduced_parser_value2, 0, sizeof(reduced_parser_value2));
             if (parser_user_code(&reduced_parser_value2, parser_reduce_table[reduce_index].rule, &statevalues, parser_reduce_table[reduce_index].n_states, context) == P_USER_TERMINATED)
             {
+                state_values_stack_free(&statevalues);
                 return P_USER_TERMINATED;
             }
             reduced_parser_value = reduced_parser_value2;
+            if (parser_reduce_table[reduce_index].n_states > 0u)
+            {
+                reduced_position = get_rule_position(&statevalues, 0u, parser_reduce_table[reduce_index].n_states, false);
+                reduced_end_position = get_rule_position(&statevalues, 0u, parser_reduce_table[reduce_index].n_states, true);
+            }
+            else
+            {
+                memset(&reduced_position, 0, sizeof(reduced_position));
+                memset(&reduced_end_position, 0, sizeof(reduced_end_position));
+            }
 <% end %>
             reduced_rule_set = parser_reduce_table[reduce_index].rule_set;
             state_values_stack_pop(&statevalues, parser_reduce_table[reduce_index].n_states);
@@ -1082,6 +1217,19 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
     return result;
 }
 
+size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
+{
+    return parse_from(context, 0u);
+}
+
+<% @grammar.start_rules.each_with_index do |start_rule, i| %>
+
+size_t <%= @grammar.prefix %>parse_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return parse_from(context, <%= i %>u);
+}
+<% end %>
+
 /**
  * Get the parse result value.
  *
@@ -1090,18 +1238,29 @@ size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
  *
  * @return Parse result value.
  */
-<% if @grammar.ast %>
-<%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
+<% if @grammar.tree %>
+<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
+{
+    return (<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> *) context->parse_result;
+}
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return (<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> *) context->parse_result;
+}
+<%   end %>
 <% else %>
 <%= start_rule_type[1] %> <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
-<% end %>
 {
-<% if @grammar.ast %>
-    return context->parse_result;
-<% else %>
     return context->parse_result.v_<%= start_rule_type[0] %>;
-<% end %>
 }
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+<%= start_rule_type(i)[1] %> <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return context->parse_result.v_<%= start_rule_type(i)[0] %>;
+}
+<%   end %>
+<% end %>
 
 /**
  * Get the current text input position.
@@ -1138,3 +1297,48 @@ size_t <%= @grammar.prefix %>user_terminate_code(<%= @grammar.prefix %>context_t
 {
     return context->token;
 }
+<% if @grammar.tree %>
+
+static void tree_delete(TreeNode * node)
+{
+    if (node->is_token)
+    {
+        <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> * token_tree_node = (<%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> *)node;
+<%= expand_code(@grammar.free_token_node, false, nil, nil) %>
+<%   if @cpp %>
+        delete token_tree_node;
+<%   else %>
+        free(token_tree_node);
+<%   end %>
+    }
+    else if (node->n_fields > 0u)
+    {
+        for (size_t i = 0u; i < node->n_fields; i++)
+        {
+            if (node->fields[i] != NULL)
+            {
+                tree_delete(node->fields[i]);
+            }
+        }
+        free(node);
+    }
+}
+
+/**
+ * Free all tree node memory.
+ */
+void <%= @grammar.prefix %>tree_delete(<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * tree)
+{
+    tree_delete((TreeNode *)tree);
+}
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+
+/**
+ * Free all tree node memory.
+ */
+void <%= @grammar.prefix %>tree_delete_<%= start_rule %>(<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * tree)
+{
+    tree_delete((TreeNode *)tree);
+}
+<%   end %>
+<% end %>

assets/parser.d.erb

@@ -8,7 +8,8 @@
 module <%= @grammar.modulename %>;
 <% end %>
 
-import core.stdc.stdlib : malloc;
+import core.memory;
+import core.stdc.stdlib : malloc, free;
 
 /**************************************************************************
  * User code blocks
@@ -58,10 +59,10 @@ public alias <%= @grammar.prefix %>code_point_t = uint;
  */
 public struct <%= @grammar.prefix %>position_t
 {
-    /** Input text row (0-based). */
+    /** Input text row (1-based). */
     uint row;
 
-    /** Input text column (0-based). */
+    /** Input text column (1-based). */
     uint col;
 
     /** Invalid position value. */
@@ -74,7 +75,7 @@ public struct <%= @grammar.prefix %>position_t
     }
 }
 
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 /** Parser values type. */
 public alias <%= @grammar.prefix %>value_t = <%= @grammar.ptype %>;
 <% else %>
@@ -85,35 +86,58 @@ public union <%= @grammar.prefix %>value_t
     <%= typestring %> v_<%= name %>;
 <%   end %>
 }
+
+/** Parser value constructor(s). */
+<%   @grammar.ptypes.each do |name, typestring| %>
+public <%= @grammar.prefix %>value_t <%= @grammar.prefix %>value<%= name == "default" ? "" : "_#{name}" %>(T)(T v)
+{
+    return <%= @grammar.prefix %>value_t(v_<%= name %>: v);
+}
 <%   end %>
 
-<% if @grammar.ast %>
-/** Common AST node structure. */
-private struct ASTNode
+/** Parser value accessor(s). */
+<%   @grammar.ptypes.each do |name, typestring| %>
+public <%= typestring %> <%= @grammar.prefix %>value_get<%= name == "default" ? "" : "_#{name}" %>(<%= @grammar.prefix %>value_t * pvalue)
+{
+    return pvalue.v_<%= name %>;
+}
+<%   end %>
+<% end %>
+
+<% if @grammar.tree %>
+/** Common tree node structure. */
+private struct TreeNode
 {
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
+    ushort n_fields;
+    bool is_token;
     void *[0] fields;
 }
 
-/** AST node types. @{ */
-public struct <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %>
+/** Tree node types. @{ */
+public struct <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>
 {
-    /* ASTNode fields must be present in the same order here. */
+    /* TreeNode fields must be present in the same order here. */
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
+    ushort n_fields;
+    bool is_token;
     <%= @grammar.prefix %>token_t token;
     <%= @grammar.prefix %>value_t pvalue;
+<%= @grammar.token_user_fields %>
 }
 
 <%   @parser.rule_sets.each do |name, rule_set| %>
 <%     next if name.start_with?("$") %>
 <%     next if rule_set.optional? %>
-public struct <%= @grammar.ast_prefix %><%= name %><%= @grammar.ast_suffix %>
+public struct <%= @grammar.tree_prefix %><%= name %><%= @grammar.tree_suffix %>
 {
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
-<%     rule_set.ast_fields.each do |fields| %>
+    ushort n_fields;
+    bool is_token;
+<%     rule_set.tree_fields.each do |fields| %>
     union
     {
 <%       fields.each do |field_name, type| %>
@@ -171,8 +195,8 @@ public struct <%= @grammar.prefix %>context_t
     /* Parser context data. */
 
     /** Parse result value. */
-<% if @grammar.ast %>
-    <%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> * parse_result;
+<% if @grammar.tree %>
+    void * parse_result;
 <% else %>
     <%= @grammar.prefix %>value_t parse_result;
 <% end %>
@@ -182,6 +206,8 @@ public struct <%= @grammar.prefix %>context_t
 
     /** User terminate code. */
     size_t user_terminate_code;
+
+<%= @grammar.context_user_fields %>
 }
 
 /**************************************************************************
@@ -221,26 +247,39 @@ private enum size_t INVALID_ID = cast(size_t)-1;
  *************************************************************************/
 
 /**
- * Initialize lexer/parser context structure.
+ * Allocate and initialize lexer/parser context structure.
+ *
+ * Deinitialize and deallocate with <%= @grammar.prefix %>context_delete().
  *
- * @param[out] context
- *   Lexer/parser context structure.
  * @param input
  *   Text input.
+ * @param input_length
+ *   Text input length.
+ *
+ * @return Context structure for lexer/parser.
  */
-public void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, string input)
+<%= @grammar.prefix %>context_t * <%= @grammar.prefix %>context_new(string input)
 {
     /* New default-initialized context structure. */
-    <%= @grammar.prefix %>context_t newcontext;
+    <%= @grammar.prefix %>context_t * context = new <%= @grammar.prefix %>context_t;
 
     /* Lexer initialization. */
-    newcontext.input = input;
-    newcontext.text_position.row = 1u;
-    newcontext.text_position.col = 1u;
-    newcontext.mode = <%= @lexer.mode_id("default") %>;
+    context.input = input;
+    context.text_position.row = 1u;
+    context.text_position.col = 1u;
+    context.mode = <%= @lexer.mode_id("default") %>;
 
-    /* Copy to the user's context structure. */
-    *context = newcontext;
+    return context;
+}
+
+/**
+ * Deinitialize and deallocate lexer/parser context structure.
+ *
+ * @param context
+ *   Lexer/parser context structure allocated with <%= @grammar.prefix %>context_new().
+ */
+void <%= @grammar.prefix %>context_delete(<%= @grammar.prefix %>context_t * context)
+{
 }
 
 /**************************************************************************
@@ -626,6 +665,22 @@ private size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%=
     {
     case P_SUCCESS:
         <%= @grammar.prefix %>token_t token_to_accept = match_info.accepting_state.token;
+        /* Calculate the token length and start/end positions before invoking
+         * the lexer user code so that the user code can access them. The
+         * context input text position tracking is not updated until after the
+         * user code has run so that it is left unchanged if the user code
+         * requests to terminate the lexer. */
+        token_info.length = match_info.length;
+        if (match_info.end_delta_position.row != 0u)
+        {
+            token_info.end_position.row = token_info.position.row + match_info.end_delta_position.row;
+            token_info.end_position.col = match_info.end_delta_position.col;
+        }
+        else
+        {
+            token_info.end_position.row = token_info.position.row;
+            token_info.end_position.col = token_info.position.col + match_info.end_delta_position.col;
+        }
         if (match_info.accepting_state.code_id != INVALID_USER_CODE_ID)
         {
             string match = context.input[context.input_index..(context.input_index + match_info.length)];
@@ -664,17 +719,6 @@ private size_t attempt_lex_token(<%= @grammar.prefix %>context_t * context, <%=
             return P_DROP;
         }
         token_info.token = token_to_accept;
-        token_info.length = match_info.length;
-        if (match_info.end_delta_position.row != 0u)
-        {
-            token_info.end_position.row = token_info.position.row + match_info.end_delta_position.row;
-            token_info.end_position.col = match_info.end_delta_position.col;
-        }
-        else
-        {
-            token_info.end_position.row = token_info.position.row;
-            token_info.end_position.col = token_info.position.col + match_info.end_delta_position.col;
-        }
         *out_token_info = token_info;
         return P_SUCCESS;
 
@@ -796,7 +840,7 @@ private struct reduce_t
      * reduce action.
      */
     parser_state_id_t n_states;
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 
     /**
      * Map of rule components to rule set child fields.
@@ -804,7 +848,7 @@ private struct reduce_t
     immutable(ushort) * rule_set_node_field_index_map;
 
     /**
-     * Number of rule set AST node fields.
+     * Number of rule set tree node fields.
      */
     ushort rule_set_node_field_array_size;
 
@@ -843,12 +887,14 @@ private struct state_value_t
     /** Parser state ID. */
     size_t state_id;
 
+<% if @grammar.tree %>
+    /** Tree node. */
+    void * tree_node;
+<% else %>
+    <%= @grammar.prefix %>position_t position;
+    <%= @grammar.prefix %>position_t end_position;
     /** Parser value from this state. */
     <%= @grammar.prefix %>value_t pvalue;
-
-<% if @grammar.ast %>
-    /** AST node. */
-    void * ast_node;
 <% end %>
 
     this(size_t state_id)
@@ -864,7 +910,7 @@ private immutable shift_t[] parser_shift_table = [
 <%   end %>
 ];
 
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 <%   @grammar.rules.each do |rule| %>
 <%     unless rule.flat_rule_set_node_field_index_map? %>
 immutable ushort[<%= rule.rule_set_node_field_index_map.size %>] r_<%= rule.name.gsub("$", "_") %><%= rule.id %>_node_field_index_map = [<%= rule.rule_set_node_field_index_map.map {|v| v.to_s}.join(", ") %>];
@@ -875,17 +921,22 @@ immutable ushort[<%= rule.rule_set_node_field_index_map.size %>] r_<%= rule.name
 /** Parser reduce table. */
 private immutable reduce_t[] parser_reduce_table = [
 <%   @parser.reduce_table.each do |reduce| %>
-    reduce_t(<%= reduce[:token_id] %>u, <%= reduce[:rule_id] %>u, <%= reduce[:rule_set_id] %>u, <%= reduce[:n_states] %>u
-<%     if @grammar.ast %>
+    reduce_t(
+        <%= reduce[:token_id] %>u, /* Token: <%= reduce[:token] ? reduce[:token].name : "(any)" %> */
+        <%= reduce[:rule_id] %>u, /* Rule ID */
+        <%= reduce[:rule_set_id] %>u, /* Rule set ID (<%= reduce[:rule].rule_set.name %>) */
+<%     if @grammar.tree %>
+        <%= reduce[:n_states] %>u, /* Number of states */
 <%       if reduce[:rule].flat_rule_set_node_field_index_map? %>
-             , null
+        null, /* No rule set node field index map (flat map) */
 <%       else %>
-             , &r_<%= reduce[:rule].name.gsub("$", "_") %><%= reduce[:rule].id %>_node_field_index_map[0]
+        &r_<%= reduce[:rule].name.gsub("$", "_") %><%= reduce[:rule].id %>_node_field_index_map[0], /* Rule set node field index map */
 <%       end %>
-             , <%= reduce[:rule].rule_set.ast_fields.size %>
-             , <%= reduce[:propagate_optional_target] %>
+        <%= reduce[:rule].rule_set.tree_fields.size %>, /* Number of tree fields */
+        <%= reduce[:propagate_optional_target] %>), /* Propagate optional target? */
+<%     else %>
+        <%= reduce[:n_states] %>u), /* Number of states */
 <%     end %>
-            ),
 <%   end %>
 ];
 
@@ -896,7 +947,56 @@ private immutable parser_state_t[] parser_state_table = [
 <%   end %>
 ];
 
-<% unless @grammar.ast %>
+<% unless @grammar.tree %>
+/**
+ * Get the rule position (start or end) for the currently matched rule.
+ */
+private <%= @grammar.prefix %>position_t get_rule_position(state_value_t[] statevalues, size_t i, size_t n_states, bool get_end)
+{
+    if (n_states > 0u)
+    {
+        if (i == 0u)
+        {
+            if (get_end)
+            {
+                for (size_t j = 0u; j < n_states; j++)
+                {
+                    state_value_t * sv = &statevalues[$-1-j];
+                    if (sv.end_position.valid)
+                    {
+                        return sv.end_position;
+                    }
+                }
+            }
+            else
+            {
+                for (size_t j = 0u; j < n_states; j++)
+                {
+                    state_value_t * sv = &statevalues[$-n_states+j];
+                    if (sv.position.valid)
+                    {
+                        return sv.position;
+                    }
+                }
+            }
+        }
+        else
+        {
+            if (get_end)
+            {
+                return statevalues[$-1-n_states+i].end_position;
+            }
+            else
+            {
+                return statevalues[$-1-n_states+i].position;
+            }
+        }
+    }
+    return <%= @grammar.prefix %>position_t.INVALID;
+}
+<% end %>
+
+<% if !@grammar.tree || @grammar.parser_user_code_used? %>
 /**
  * Execute user code associated with a parser rule.
  *
@@ -907,7 +1007,7 @@ private immutable parser_state_t[] parser_state_table = [
  * @retval P_USER_TERMINATED
  *   User requested to terminate parsing.
  */
-private size_t parser_user_code(<%= @grammar.prefix %>value_t * _pvalue, uint rule, state_value_t[] statevalues, uint n_states, <%= @grammar.prefix %>context_t * context)
+private size_t parser_user_code(<%= @grammar.tree ? "void" : "#{@grammar.prefix}value_t" %> * _pvalue, uint rule, state_value_t[] statevalues, uint n_states, <%= @grammar.prefix %>context_t * context)
 {
     switch (rule)
     {
@@ -979,6 +1079,8 @@ private size_t check_reduce(size_t state_id, <%= @grammar.prefix %>token_t token
  *
  * @param context
  *   Lexer/parser context structure.
+ * @start_state_id
+ *   ID of the state in which to start.
  *
  * @retval P_SUCCESS
  *   The parser successfully matched the input text. The parse result value
@@ -991,22 +1093,25 @@ private size_t check_reduce(size_t state_id, <%= @grammar.prefix %>token_t token
  * @reval P_UNEXPECTED_INPUT
  *   Input text does not match any lexer pattern.
  */
-public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
+private size_t parse_from(<%= @grammar.prefix %>context_t * context, size_t start_state_id)
 {
     <%= @grammar.prefix %>token_info_t token_info;
     <%= @grammar.prefix %>token_t token = INVALID_TOKEN_ID;
     state_value_t[] statevalues = new state_value_t[](1);
+    statevalues[0].state_id = start_state_id;
     size_t reduced_rule_set = INVALID_ID;
-<% if @grammar.ast %>
+<% if @grammar.tree %>
     void * reduced_parser_node;
 <% else %>
+    <%= @grammar.prefix %>position_t reduced_position;
+    <%= @grammar.prefix %>position_t reduced_end_position;
     <%= @grammar.prefix %>value_t reduced_parser_value;
 <% end %>
     for (;;)
     {
         if (token == INVALID_TOKEN_ID)
         {
-            size_t lexer_result = <%= @grammar.prefix %>lex(context, &token_info);
+            size_t lexer_result = <%= lex_fn %>(context, &token_info);
             if (lexer_result != P_SUCCESS)
             {
                 return lexer_result;
@@ -1024,8 +1129,8 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
             if ((shift_state != INVALID_ID) && (token == TOKEN___EOF))
             {
                 /* Successful parse. */
-<% if @grammar.ast %>
-                context.parse_result = cast(<%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> *)statevalues[$-1].ast_node;
+<% if @grammar.tree %>
+                context.parse_result = statevalues[$-1].tree_node;
 <% else %>
                 context.parse_result = statevalues[$-1].pvalue;
 <% end %>
@@ -1039,10 +1144,13 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
             if (reduced_rule_set == INVALID_ID)
             {
                 /* We shifted a token, mark it consumed. */
-<% if @grammar.ast %>
-                <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %> * token_ast_node = new <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %>(token_info.position, token_info.end_position, token, token_info.pvalue);
-                statevalues[$-1].ast_node = token_ast_node;
+<% if @grammar.tree %>
+                <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %> * token_tree_node = new <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>(token_info.position, token_info.end_position, 0u, true, token, token_info.pvalue);
+<%= expand_code(@grammar.on_token_node, false, nil, nil) %>
+                statevalues[$-1].tree_node = token_tree_node;
 <% else %>
+                statevalues[$-1].position = token_info.position;
+                statevalues[$-1].end_position = token_info.end_position;
                 statevalues[$-1].pvalue = token_info.pvalue;
 <% end %>
                 token = INVALID_TOKEN_ID;
@@ -1050,10 +1158,12 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
             else
             {
                 /* We shifted a RuleSet. */
-<% if @grammar.ast %>
-                statevalues[$-1].ast_node = reduced_parser_node;
+<% if @grammar.tree %>
+                statevalues[$-1].tree_node = reduced_parser_node;
 <% else %>
                 statevalues[$-1].pvalue = reduced_parser_value;
+                statevalues[$-1].position = reduced_position;
+                statevalues[$-1].end_position = reduced_end_position;
                 <%= @grammar.prefix %>value_t new_parse_result;
                 reduced_parser_value = new_parse_result;
 <% end %>
@@ -1066,17 +1176,21 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
         if (reduce_index != INVALID_ID)
         {
             /* We have something to reduce. */
-<% if @grammar.ast %>
+<% if @grammar.tree %>
             if (parser_reduce_table[reduce_index].propagate_optional_target)
             {
-                reduced_parser_node = statevalues[$ - 1].ast_node;
+                reduced_parser_node = statevalues[$ - 1].tree_node;
             }
             else if (parser_reduce_table[reduce_index].n_states > 0)
             {
                 size_t n_fields = parser_reduce_table[reduce_index].rule_set_node_field_array_size;
-                ASTNode * node = cast(ASTNode *)malloc(ASTNode.sizeof + n_fields * (void *).sizeof);
+                size_t node_size = TreeNode.sizeof + n_fields * (void *).sizeof;
+                TreeNode * node = cast(TreeNode *)malloc(node_size);
+                GC.addRange(node, node_size);
                 node.position = <%= @grammar.prefix %>position_t.INVALID;
                 node.end_position = <%= @grammar.prefix %>position_t.INVALID;
+                node.n_fields = cast(ushort)n_fields;
+                node.is_token = false;
                 foreach (i; 0..n_fields)
                 {
                     node.fields[i] = null;
@@ -1085,20 +1199,20 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
                 {
                     foreach (i; 0..parser_reduce_table[reduce_index].n_states)
                     {
-                        node.fields[i] = statevalues[$ - parser_reduce_table[reduce_index].n_states + i].ast_node;
+                        node.fields[i] = statevalues[$ - parser_reduce_table[reduce_index].n_states + i].tree_node;
                     }
                 }
                 else
                 {
                     foreach (i; 0..parser_reduce_table[reduce_index].n_states)
                     {
-                        node.fields[parser_reduce_table[reduce_index].rule_set_node_field_index_map[i]] = statevalues[$ - parser_reduce_table[reduce_index].n_states + i].ast_node;
+                        node.fields[parser_reduce_table[reduce_index].rule_set_node_field_index_map[i]] = statevalues[$ - parser_reduce_table[reduce_index].n_states + i].tree_node;
                     }
                 }
                 bool position_found = false;
                 foreach (i; 0..n_fields)
                 {
-                    ASTNode * child = cast(ASTNode *)node.fields[i];
+                    TreeNode * child = cast(TreeNode *)node.fields[i];
                     if (child && child.position.valid)
                     {
                         if (!position_found)
@@ -1115,6 +1229,12 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
             {
                 reduced_parser_node = null;
             }
+<%   if @grammar.parser_user_code_used? %>
+            if (parser_user_code(reduced_parser_node, parser_reduce_table[reduce_index].rule, statevalues, parser_reduce_table[reduce_index].n_states, context) == P_USER_TERMINATED)
+            {
+                return P_USER_TERMINATED;
+            }
+<%   end %>
 <% else %>
             <%= @grammar.prefix %>value_t reduced_parser_value2;
             if (parser_user_code(&reduced_parser_value2, parser_reduce_table[reduce_index].rule, statevalues, parser_reduce_table[reduce_index].n_states, context) == P_USER_TERMINATED)
@@ -1122,6 +1242,16 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
                 return P_USER_TERMINATED;
             }
             reduced_parser_value = reduced_parser_value2;
+            if (parser_reduce_table[reduce_index].n_states > 0u)
+            {
+                reduced_position = get_rule_position(statevalues, 0u, parser_reduce_table[reduce_index].n_states, false);
+                reduced_end_position = get_rule_position(statevalues, 0u, parser_reduce_table[reduce_index].n_states, true);
+            }
+            else
+            {
+                reduced_position = <%= @grammar.prefix %>position_t.INVALID;
+                reduced_end_position = <%= @grammar.prefix %>position_t.INVALID;
+            }
 <% end %>
             reduced_rule_set = parser_reduce_table[reduce_index].rule_set;
             statevalues.length -= parser_reduce_table[reduce_index].n_states;
@@ -1139,6 +1269,19 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
     }
 }
 
+public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context)
+{
+    return parse_from(context, 0u);
+}
+
+<% @grammar.start_rules.each_with_index do |start_rule, i| %>
+
+public size_t <%= @grammar.prefix %>parse_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return parse_from(context, <%= i %>u);
+}
+<% end %>
+
 /**
  * Get the parse result value.
  *
@@ -1147,18 +1290,58 @@ public size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * cont
  *
  * @return Parse result value.
  */
-<% if @grammar.ast %>
-public <%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
+<% if @grammar.tree %>
+public <%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
+{
+    return cast(<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> *)context.parse_result;
+}
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+public <%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return cast(<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> *)context.parse_result;
+}
+<%   end %>
 <% else %>
 public <%= start_rule_type[1] %> <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context)
-<% end %>
 {
-<% if @grammar.ast %>
-    return context.parse_result;
-<% else %>
     return context.parse_result.v_<%= start_rule_type[0] %>;
-<% end %>
 }
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+public <%= start_rule_type(i)[1] %> <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context)
+{
+    return context.parse_result.v_<%= start_rule_type(i)[0] %>;
+}
+<%   end %>
+<% end %>
+
+<% if @grammar.tree %>
+private void tree_delete(TreeNode * node)
+{
+    if (!node.is_token)
+    {
+        for (size_t i = 0u; i < node.n_fields; i++)
+        {
+            if (node.fields[i])
+            {
+                tree_delete(cast(TreeNode *)node.fields[i]);
+            }
+        }
+        GC.removeRange(node);
+        free(node);
+    }
+}
+
+void <%= @grammar.prefix %>tree_delete(<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * tree)
+{
+    tree_delete(cast(TreeNode *)tree);
+}
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+void <%= @grammar.prefix %>tree_delete_<%= start_rule %>(<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * tree)
+{
+    tree_delete(cast(TreeNode *)tree);
+}
+<%   end %>
+<% end %>
 
 /**
  * Get the current text input position.

assets/parser.h.erb

@@ -45,10 +45,10 @@ typedef uint32_t <%= @grammar.prefix %>code_point_t;
  */
 typedef struct
 {
-    /** Input text row (0-based). */
+    /** Input text row (1-based). */
     uint32_t row;
 
-    /** Input text column (0-based). */
+    /** Input text column (1-based). */
     uint32_t col;
 } <%= @grammar.prefix %>position_t;
 
@@ -58,7 +58,7 @@ typedef struct
 /** User header code blocks. */
 <%= @grammar.code_blocks.fetch("header", "") %>
 
-<% if @grammar.ast %>
+<% if @grammar.tree %>
 /** Parser values type. */
 typedef <%= @grammar.ptype %> <%= @grammar.prefix %>value_t;
 <% else %>
@@ -69,18 +69,37 @@ typedef union
     <%= typestring %> v_<%= name %>;
 <%   end %>
 } <%= @grammar.prefix %>value_t;
+
+/** Parser value constructor(s). */
+<%   @grammar.ptypes.each do |name, typestring| %>
+static inline <%= @grammar.prefix %>value_t <%= @grammar.prefix %>value<%= name == "default" ? "" : "_#{name}" %>(<%= typestring %> v)
+{
+    return (<%= @grammar.prefix %>value_t){.v_<%= name %> = v};
+}
 <%   end %>
 
-<% if @grammar.ast %>
-/** AST node types. @{ */
-typedef struct <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %>
+/** Parser value accessor(s). */
+<%   @grammar.ptypes.each do |name, typestring| %>
+static inline <%= typestring %> <%= @grammar.prefix %>value_get<%= name == "default" ? "" : "_#{name}" %>(<%= @grammar.prefix %>value_t const * pvalue)
 {
-    /* ASTNode fields must be present in the same order here. */
+    return pvalue->v_<%= name %>;
+}
+<%   end %>
+<% end %>
+
+<% if @grammar.tree %>
+/** Tree node types. @{ */
+typedef struct <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>
+{
+    <% # TreeNode fields must be present in the same order here. # %>
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
+    uint16_t n_fields;
+    uint8_t is_token;
+<%= @grammar.token_user_fields %>
     <%= @grammar.prefix %>token_t token;
     <%= @grammar.prefix %>value_t pvalue;
-} <%= @grammar.ast_prefix %>Token<%= @grammar.ast_suffix %>;
+} <%= @grammar.tree_prefix %>Token<%= @grammar.tree_suffix %>;
 
 <%   @parser.rule_sets.each do |name, rule_set| %>
 <%     next if name.start_with?("$") %>
@@ -91,11 +110,14 @@ struct <%= name %>;
 <%   @parser.rule_sets.each do |name, rule_set| %>
 <%     next if name.start_with?("$") %>
 <%     next if rule_set.optional? %>
-typedef struct <%= @grammar.ast_prefix %><%= name %><%= @grammar.ast_suffix %>
+typedef struct <%= @grammar.tree_prefix %><%= name %><%= @grammar.tree_suffix %>
 {
+    <% # TreeNode fields must be present in the same order here. # %>
     <%= @grammar.prefix %>position_t position;
     <%= @grammar.prefix %>position_t end_position;
-<%     rule_set.ast_fields.each do |fields| %>
+    uint16_t n_fields;
+    uint8_t is_token;
+<%     rule_set.tree_fields.each do |fields| %>
     union
     {
 <%       fields.each do |field_name, type| %>
@@ -103,7 +125,7 @@ typedef struct <%= @grammar.ast_prefix %><%= name %><%= @grammar.ast_suffix %>
 <%       end %>
     };
 <%     end %>
-} <%= @grammar.ast_prefix %><%= name %><%= @grammar.ast_suffix %>;
+} <%= @grammar.tree_prefix %><%= name %><%= @grammar.tree_suffix %>;
 
 <%   end %>
 /** @} */
@@ -156,8 +178,8 @@ typedef struct
     /* Parser context data. */
 
     /** Parse result value. */
-<% if @grammar.ast %>
-    <%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> * parse_result;
+<% if @grammar.tree %>
+    void * parse_result;
 <% else %>
     <%= @grammar.prefix %>value_t parse_result;
 <% end %>
@@ -167,6 +189,8 @@ typedef struct
 
     /** User terminate code. */
     size_t user_terminate_code;
+
+<%= @grammar.context_user_fields %>
 } <%= @grammar.prefix %>context_t;
 
 /**************************************************************************
@@ -176,7 +200,9 @@ typedef struct
 /** Token names. */
 extern const char * <%= @grammar.prefix %>token_names[];
 
-void <%= @grammar.prefix %>context_init(<%= @grammar.prefix %>context_t * context, uint8_t const * input, size_t input_length);
+<%= @grammar.prefix %>context_t * <%= @grammar.prefix %>context_new(uint8_t const * input, size_t input_length);
+
+void <%= @grammar.prefix %>context_delete(<%= @grammar.prefix %>context_t * context);
 
 size_t <%= @grammar.prefix %>decode_code_point(uint8_t const * input, size_t input_length,
         <%= @grammar.prefix %>code_point_t * out_code_point, uint8_t * out_code_point_length);
@@ -184,11 +210,27 @@ size_t <%= @grammar.prefix %>decode_code_point(uint8_t const * input, size_t inp
 size_t <%= @grammar.prefix %>lex(<%= @grammar.prefix %>context_t * context, <%= @grammar.prefix %>token_info_t * out_token_info);
 
 size_t <%= @grammar.prefix %>parse(<%= @grammar.prefix %>context_t * context);
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+size_t <%= @grammar.prefix %>parse_<%= start_rule %>(<%= @grammar.prefix %>context_t * context);
+<%   end %>
 
-<% if @grammar.ast %>
-<%= @grammar.ast_prefix %><%= @grammar.start_rule %><%= @grammar.ast_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context);
+<% if @grammar.tree %>
+<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context);
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context);
+<%   end %>
 <% else %>
 <%= start_rule_type[1] %> <%= @grammar.prefix %>result(<%= @grammar.prefix %>context_t * context);
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+<%= start_rule_type(i)[1] %> <%= @grammar.prefix %>result_<%= start_rule %>(<%= @grammar.prefix %>context_t * context);
+<%   end %>
+<% end %>
+
+<% if @grammar.tree %>
+void <%= @grammar.prefix %>tree_delete(<%= @grammar.tree_prefix %><%= @grammar.start_rules[0] %><%= @grammar.tree_suffix %> * tree);
+<%   @grammar.start_rules.each_with_index do |start_rule, i| %>
+void <%= @grammar.prefix %>tree_delete_<%= start_rule %>(<%= @grammar.tree_prefix %><%= start_rule %><%= @grammar.tree_suffix %> * tree);
+<%   end %>
 <% end %>
 
 <%= @grammar.prefix %>position_t <%= @grammar.prefix %>position(<%= @grammar.prefix %>context_t * context);

extra/vim/syntax/propane.vim

@@ -8,7 +8,11 @@ if exists("b:current_syntax")
 endif
 
 if !exists("b:propane_subtype")
+  if search('\<import\s\+\%(std\|core\)\.', 'nw') > 0
     let b:propane_subtype = "d"
+  else
+    let b:propane_subtype = "cpp"
+  endif
 endif
 
 exe "syn include @propaneTarget syntax/".b:propane_subtype.".vim"
@@ -16,18 +20,32 @@ exe "syn include @propaneTarget syntax/".b:propane_subtype.".vim"
 syn region propaneTarget matchgroup=propaneDelimiter start="<<" end=">>$" contains=@propaneTarget keepend
 
 syn match propaneComment "#.*"
-syn match propaneOperator "->"
 syn match propaneFieldAlias ":[a-zA-Z0-9_]\+" contains=propaneFieldOperator
 syn match propaneFieldOperator ":" contained
 syn match propaneOperator "?"
-syn keyword propaneKeyword ast ast_prefix ast_suffix drop module prefix ptype start token tokenid
+" The right-hand side of a rule (after '->' up to '<<' or ';') lists symbol
+" names that may coincide with propane keywords (e.g. 'start', 'token',
+" 'tree'). Wrap it in a region that excludes keyword matches so those names
+" are not highlighted as keywords. The '<<' is left unconsumed so the
+" propaneTarget region can still match it.
+syn region propaneRuleRhs matchgroup=propaneOperator start="->" end="\ze<<" end=";" contains=propaneFieldAlias,propaneRuleOperator,propaneComment keepend
+syn match propaneRuleOperator "?" contained
+" Keywords that introduce a user-defined name. The name is consumed by
+" propaneName via nextgroup so a name matching a keyword (e.g. 'token start')
+" is not highlighted as a keyword. These must be a match (not syn keyword)
+" because a syn keyword always wins over a contained nextgroup match.
+syn match propaneNameDecl "\<\%(tokenid\|token\|lex_fn\|module\|start\|tree_prefix\|tree_suffix\)\>" nextgroup=propaneName skipwhite
+syn match propaneName "\<\h\w*\>" contained
+syn match propaneKeyword "\<\%(context_user_fields\|drop\|free_token_node\|noline\|on_token_node\|prefix\|ptype\|token_user_fields\|tree\)\>"
 
-syn region propaneRegex start="/" end="/" skip="\\/"
+syn region propaneRegex start="/" end="/" skip="\v\\\\|\\/"
 
 hi def link propaneComment Comment
 hi def link propaneKeyword Keyword
+hi def link propaneNameDecl Keyword
 hi def link propaneRegex String
 hi def link propaneOperator Operator
+hi def link propaneRuleOperator Operator
 hi def link propaneFieldOperator Operator
 hi def link propaneDelimiter Delimiter
 hi def link propaneFieldAlias Identifier

lib/propane.rb

@@ -33,7 +33,7 @@ class Propane
 
     def run(input_file, output_file, log_file, options)
       begin
-        grammar = Grammar.new(File.read(input_file))
+        grammar = Grammar.new(File.read(input_file), input_file)
         generator = Generator.new(grammar, output_file, log_file, options)
         generator.generate
       rescue Error => e

lib/propane/generator.rb

@@ -11,25 +11,40 @@ class Propane
         @log = StringIO.new
       end
       @language =
-        if output_file =~ /\.([a-z]+)$/
-          $1
-        else
+        if output_file.end_with?(".d")
           "d"
+        elsif output_file.end_with?(".c")
+          "c"
+        elsif output_file =~ %r{\.(cc|cpp|cxx)$}
+          @cpp = true
+          "c"
+        else
+          raise Error.new("Could not determine target language from output file name (#{output_file})")
         end
       @options = options
       process_grammar!
     end
 
     def generate
-      extensions = [@language]
+      extensions = [nil]
       if @language == "c"
         extensions += %w[h]
       end
       extensions.each do |extension|
-        template = Assets.get("parser.#{extension}.erb")
-        erb = ERB.new(template, trim_mode: "<>")
+        template = Assets.get("parser.#{extension || @language}.erb")
+        if extension
           output_file = @output_file.sub(%r{\.[a-z]+$}, ".#{extension}")
-        result = erb.result(binding.clone)
+        else
+          output_file = @output_file
+        end
+        erb = ERB.new(template, trim_mode: "<>")
+        result = erb.result(binding.clone).lines.each_with_index.map do |line, i|
+          if line == "#linereset\n"
+            %[#line #{i + 2} "#{output_file}"\n]
+          else
+            line
+          end
+        end.join
         File.open(output_file, "wb") do |fh|
           fh.write(result)
         end
@@ -67,12 +82,15 @@ class Propane
         end
         tokens_by_name[token.name] = token
       end
-      # Check for user start rule.
-      unless @grammar.rules.find {|rule| rule.name == @grammar.start_rule}
-        raise Error.new("Start rule `#{@grammar.start_rule}` not found")
+      # Create real start rule(s).
+      real_start_rules = @grammar.start_rules.map do |start_rule|
+        unless @grammar.rules.find {|rule| rule.name == start_rule}
+          raise Error.new("Start rule `#{start_rule}` not found")
         end
-      # Add "real" start rule.
-      @grammar.rules.unshift(Rule.new("$Start", [@grammar.start_rule, "$EOF"], nil, nil, nil))
+        Rule.new("$#{start_rule}", [start_rule, "$EOF"], nil, nil, nil)
+      end
+      # Add real start rules before user-given rules.
+      @grammar.rules = real_start_rules + @grammar.rules
       # Generate and add rules for optional components.
       generate_optional_component_rules!(tokens_by_name)
       # Build rule sets.
@@ -167,7 +185,8 @@ class Propane
                 end
               end
               @grammar.rules << Rule.new(component, [], nil, ptypename, rule.line_number)
-              @grammar.rules << Rule.new(component, [c], "$$ = $1;\n", ptypename, rule.line_number)
+              optcode = @grammar.tree ? nil : "$$ = $1;\n"
+              @grammar.rules << Rule.new(component, [c], optcode, ptypename, rule.line_number)
               optional_rules_added << component
             end
           end
@@ -258,8 +277,34 @@ class Propane
           "context.user_terminate_code = (#{user_terminate_code}); return #{retval};"
         end
       end
+      code = code.gsub(/\$\{context\.(\w+)\}/) do |match|
+        fieldname = $1
+        case @language
+        when "c"
+          "context->#{fieldname}"
+        when "d"
+          "context.#{fieldname}"
+        end
+      end
+      code = code.gsub(/\$\{token\.(\w+)\}/) do |match|
+        fieldname = $1
+        case @language
+        when "c"
+          "token_tree_node->#{fieldname}"
+        when "d"
+          "token_tree_node.#{fieldname}"
+        end
+      end
       if parser
         code = code.gsub(/\$\$/) do |match|
+          if @grammar.tree
+            case @language
+            when "c"
+              "((#{@grammar.tree_prefix}#{rule.name}#{@grammar.tree_suffix} *)_pvalue)"
+            when "d"
+              "(cast(#{@grammar.tree_prefix}#{rule.name}#{@grammar.tree_suffix} *)_pvalue)"
+            end
+          else
             case @language
             when "c"
               "_pvalue->v_#{rule.ptypename}"
@@ -267,31 +312,33 @@ class Propane
               "_pvalue.v_#{rule.ptypename}"
             end
           end
-        code = code.gsub(/\$(\d+)/) do |match|
-          index = $1.to_i
-          case @language
-          when "c"
-            "state_values_stack_index(statevalues, -1 - (int)n_states + #{index})->pvalue.v_#{rule.components[index - 1].ptypename}"
-          when "d"
-            "statevalues[$-1-n_states+#{index}].pvalue.v_#{rule.components[index - 1].ptypename}"
         end
+        code = code.gsub(/\$(\d+)/) do |match|
+          parser_component_reference(rule, $1.to_i)
+        end
+        code = code.gsub(/\$\{(\$|\d+)\.position\}/) do |match|
+          index = $1.to_i
+          "get_rule_position(statevalues, #{index}, n_states, false)"
+        end
+        code = code.gsub(/\$\{(\$|\d+)\.end_position\}/) do |match|
+          index = $1.to_i
+          "get_rule_position(statevalues, #{index}, n_states, true)"
         end
         code = code.gsub(/\$\{(\w+)\}/) do |match|
           aliasname = $1
           if index = rule.aliases[aliasname]
-            case @language
-            when "c"
-              "state_values_stack_index(statevalues, -(int)n_states + #{index})->pvalue.v_#{rule.components[index].ptypename}"
-            when "d"
-              "statevalues[$-n_states+#{index}].pvalue.v_#{rule.components[index].ptypename}"
-            end
+            # Field aliases are just a named reference to a positional rule
+            # component, so reuse the same expansion as `$1', `$2', etc. Note
+            # that rule.aliases stores a 0-based component index, so add 1 to
+            # convert it to the 1-based index used for positional references.
+            parser_component_reference(rule, index + 1)
           else
             raise Error.new("Field alias '#{aliasname}' not found")
           end
         end
       else
         code = code.gsub(/\$\$/) do |match|
-          if @grammar.ast
+          if @grammar.tree
             case @language
             when "c"
               "out_token_info->pvalue"
@@ -307,6 +354,22 @@ class Propane
             end
           end
         end
+        code = code.gsub(/\$\{position\}/) do |match|
+          case @language
+          when "c"
+            "out_token_info->position"
+          when "d"
+            "out_token_info.position"
+          end
+        end
+        code = code.gsub(/\$\{end_position\}/) do |match|
+          case @language
+          when "c"
+            "out_token_info->end_position"
+          when "d"
+            "out_token_info.end_position"
+          end
+        end
         code = code.gsub(/\$mode\(([a-zA-Z_][a-zA-Z_0-9]*)\)/) do |match|
           mode_name = $1
           mode_id = @lexer.mode_id(mode_name)
@@ -324,13 +387,60 @@ class Propane
       code
     end
 
+    # Expand a positional reference to a parser rule component.
+    #
+    # This is used to expand `$1', `$2', etc. as well as field aliases (which
+    # are just named references to a positional rule component).
+    #
+    # @param rule [Rule]
+    #   The Rule containing the user code.
+    # @param index [Integer]
+    #   1-based index of the rule component to reference.
+    #
+    # @return [String]
+    #   Expanded rule component reference.
+    def parser_component_reference(rule, index)
+      component = rule.components[index - 1]
+      if @grammar.tree
+        # In tree mode a component reference yields a pointer to that
+        # component's tree node. An optional component propagates its target
+        # node (or null), so use the optional target's node type.
+        if component.is_a?(RuleSet) && component.optional?
+          component = component.option_target
+        end
+        node_name = component.is_a?(Token) ? "Token" : component.name
+        typename = "#{@grammar.tree_prefix}#{node_name}#{@grammar.tree_suffix}"
+        case @language
+        when "c"
+          "((#{typename} *)state_values_stack_index(statevalues, -1 - (int)n_states + #{index})->tree_node)"
+        when "d"
+          "(cast(#{typename} *)statevalues[$-1-n_states+#{index}].tree_node)"
+        end
+      else
+        case @language
+        when "c"
+          "state_values_stack_index(statevalues, -1 - (int)n_states + #{index})->pvalue.v_#{component.ptypename}"
+        when "d"
+          "statevalues[$-1-n_states+#{index}].pvalue.v_#{component.ptypename}"
+        end
+      end
+    end
+
+    # Get the lex function to use.
+    #
+    # @return [String]
+    #   Lex function to use.
+    def lex_fn
+      @grammar.lex_fn || "#{@grammar.prefix}lex"
+    end
+
     # Get the parser value type for the start rule.
     #
     # @return [Array<String>]
     #   Start rule parser value type name and type string.
-    def start_rule_type
+    def start_rule_type(start_rule_index = 0)
       start_rule = @grammar.rules.find do |rule|
-        rule.name == @grammar.start_rule
+        rule.name == @grammar.start_rules[start_rule_index]
       end
       [start_rule.ptypename, @grammar.ptypes[start_rule.ptypename]]
     end

lib/propane/grammar.rb

@@ -5,21 +5,27 @@ class Propane
     # Reserve identifiers beginning with a double-underscore for internal use.
     IDENTIFIER_REGEX = /(?:[a-zA-Z]|_[a-zA-Z0-9])[a-zA-Z_0-9]*/
 
-    attr_reader :ast
-    attr_reader :ast_prefix
-    attr_reader :ast_suffix
+    attr_reader :context_user_fields
+    attr_reader :lex_fn
+    attr_reader :tree
+    attr_reader :tree_prefix
+    attr_reader :tree_suffix
+    attr_reader :free_token_node
     attr_reader :modulename
     attr_reader :patterns
-    attr_reader :rules
-    attr_reader :start_rule
+    attr_accessor :rules
+    attr_reader :start_rules
     attr_reader :tokens
     attr_reader :code_blocks
     attr_reader :ptypes
     attr_reader :prefix
+    attr_reader :on_token_node
+    attr_reader :token_user_fields
 
-    def initialize(input)
+    def initialize(input, filename)
+      @filename = filename
       @patterns = []
-      @start_rule = "Start"
+      @start_rules = []
       @tokens = []
       @rules = []
       @code_blocks = {}
@@ -29,10 +35,15 @@ class Propane
       @input = input.gsub("\r\n", "\n")
       @ptypes = {"default" => "void *"}
       @prefix = "p_"
-      @ast = false
-      @ast_prefix = ""
-      @ast_suffix = ""
+      @tree = false
+      @tree_prefix = ""
+      @tree_suffix = ""
+      @free_token_node = ""
+      @context_user_fields = nil
+      @on_token_node = ""
+      @token_user_fields = nil
       parse_grammar!
+      @start_rules << "Start" if @start_rules.empty?
     end
 
     def ptype
@@ -47,6 +58,10 @@ class Propane
       @tokens.size + 1
     end
 
+    def parser_user_code_used?
+      @rules.any? {|r| r.code}
+    end
+
     private
 
     def parse_grammar!
@@ -59,10 +74,15 @@ class Propane
       if parse_white_space!
       elsif parse_comment_line!
       elsif @modeline.nil? && parse_mode_label!
-      elsif parse_ast_statement!
-      elsif parse_ast_prefix_statement!
-      elsif parse_ast_suffix_statement!
+      elsif parse_context_user_fields_statement!
+      elsif parse_lex_fn!
+      elsif parse_tree_statement!
+      elsif parse_tree_prefix_statement!
+      elsif parse_tree_suffix_statement!
+      elsif parse_free_token_node_statement!
       elsif parse_module_statement!
+      elsif parse_on_token_node_statement!
+      elsif parse_token_user_fields_statement!
       elsif parse_ptype_statement!
       elsif parse_pattern_statement!
       elsif parse_start_statement!
@@ -72,6 +92,7 @@ class Propane
       elsif parse_rule_statement!
       elsif parse_code_block_statement!
       elsif parse_prefix_statement!
+      elsif parse_noline_statement!
       else
         if @input.size > 25
           @input = @input.slice(0..20) + "..."
@@ -94,21 +115,37 @@ class Propane
       consume!(/#.*\n/)
     end
 
-    def parse_ast_statement!
-      if consume!(/ast\s*;/)
-        @ast = true
+    def parse_context_user_fields_statement!
+      if md = consume!(/context_user_fields\b\s*/)
+        unless code = parse_code_block!
+          raise Error.new("Line #{@line_number}: expected code block")
+        end
+        @context_user_fields ||= ""
+        @context_user_fields += code
       end
     end
 
-    def parse_ast_prefix_statement!
-      if md = consume!(/ast_prefix\s+(\w+)\s*;/)
-        @ast_prefix = md[1]
+    def parse_lex_fn!
+      if md = consume!(/lex_fn\b\s*(\w+)\s*;/)
+        @lex_fn = md[1]
       end
     end
 
-    def parse_ast_suffix_statement!
-      if md = consume!(/ast_suffix\s+(\w+)\s*;/)
-        @ast_suffix = md[1]
+    def parse_tree_statement!
+      if consume!(/tree\s*;/)
+        @tree = true
+      end
+    end
+
+    def parse_tree_prefix_statement!
+      if md = consume!(/tree_prefix\s+(\w+)\s*;/)
+        @tree_prefix = md[1]
+      end
+    end
+
+    def parse_tree_suffix_statement!
+      if md = consume!(/tree_suffix\s+(\w+)\s*;/)
+        @tree_suffix = md[1]
       end
     end
 
@@ -122,12 +159,40 @@ class Propane
       end
     end
 
+    def parse_on_token_node_statement!
+      if md = consume!(/on_token_node\b\s*/)
+        unless code = parse_code_block!
+          raise Error.new("Line #{@line_number}: expected code block")
+        end
+        @on_token_node += code
+      end
+    end
+
+    def parse_token_user_fields_statement!
+      if md = consume!(/token_user_fields\b\s*/)
+        unless code = parse_code_block!
+          raise Error.new("Line #{@line_number}: expected code block")
+        end
+        @token_user_fields ||= ""
+        @token_user_fields += code
+      end
+    end
+
+    def parse_free_token_node_statement!
+      if md = consume!(/free_token_node\b\s*/)
+        unless code = parse_code_block!
+          raise Error.new("Line #{@line_number}: expected code block")
+        end
+        @free_token_node += code
+      end
+    end
+
     def parse_ptype_statement!
       if consume!(/ptype\s+/)
         name = "default"
         if md = consume!(/(#{IDENTIFIER_REGEX})\s*=\s*/)
-          if @ast
-            raise Error.new("Multiple ptypes are unsupported in AST mode")
+          if @tree
+            raise Error.new("Multiple ptypes are unsupported in tree mode")
           end
           name = md[1]
         end
@@ -141,8 +206,8 @@ class Propane
         md = consume!(/(#{IDENTIFIER_REGEX})\s*/, "expected token name")
         name = md[1]
         if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
-          if @ast
-            raise Error.new("Multiple ptypes are unsupported in AST mode")
+          if @tree
+            raise Error.new("Multiple ptypes are unsupported in tree mode")
           end
           ptypename = md[1]
         end
@@ -165,8 +230,8 @@ class Propane
         md = consume!(/(#{IDENTIFIER_REGEX})\s*/, "expected token name")
         name = md[1]
         if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
-          if @ast
-            raise Error.new("Multiple ptypes are unsupported in AST mode")
+          if @tree
+            raise Error.new("Multiple ptypes are unsupported in tree mode")
           end
           ptypename = md[1]
         end
@@ -185,8 +250,10 @@ class Propane
           raise Error.new("Line #{@line_number}: expected pattern to follow `drop'")
         end
         consume!(/\s+/)
-        consume!(/;/, "expected `;'")
-        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, modes: get_modes_from_modeline)
+        unless code = parse_code_block!
+          consume!(/;/, "expected `;' or code block")
+        end
+        @patterns << Pattern.new(pattern: pattern, line_number: @line_number, code: code, modes: get_modes_from_modeline)
         @modeline = nil
         true
       end
@@ -195,18 +262,14 @@ class Propane
     def parse_rule_statement!
       if md = consume!(/(#{IDENTIFIER_REGEX})\s*(?:\((#{IDENTIFIER_REGEX})\))?\s*->\s*/)
         rule_name, ptypename = *md[1, 2]
-        if @ast && ptypename
-          raise Error.new("Multiple ptypes are unsupported in AST mode")
+        if @tree && ptypename
+          raise Error.new("Multiple ptypes are unsupported in tree mode")
         end
         md = consume!(/((?:#{IDENTIFIER_REGEX}\??(?::#{IDENTIFIER_REGEX})?\s*)*)\s*/, "expected rule component list")
         components = md[1].strip.split(/\s+/)
-        if @ast
-          consume!(/;/, "expected `;'")
-        else
         unless code = parse_code_block!
           consume!(/;/, "expected `;' or code block")
         end
-        end
         @rules << Rule.new(rule_name, components, code, ptypename, @line_number)
         @modeline = nil
         true
@@ -217,8 +280,8 @@ class Propane
       if pattern = parse_pattern!
         consume!(/\s+/)
         if md = consume!(/\((#{IDENTIFIER_REGEX})\)\s*/)
-          if @ast
-            raise Error.new("Multiple ptypes are unsupported in AST mode")
+          if @tree
+            raise Error.new("Multiple ptypes are unsupported in tree mode")
           end
           ptypename = md[1]
         end
@@ -232,16 +295,25 @@ class Propane
     end
 
     def parse_start_statement!
-      if md = consume!(/start\s+(\w+)\s*;/)
-        @start_rule = md[1]
+      if md = consume!(/start\s+([\w\s]*);/)
+        start_rules = md[1].split(/\s+/).map(&:strip)
+        start_rules.each do |start_rule|
+          @start_rules << start_rule unless @start_rules.include?(start_rule)
+        end
       end
     end
 
     def parse_code_block_statement!
       if md = consume!(/<<([a-z]*)(.*?)>>\n/m)
         name, code = md[1..2]
-        code.sub!(/\A\n/, "")
-        code += "\n" unless code.end_with?("\n")
+        code = code.chomp
+        unless @noline
+          if code.start_with?("\n")
+            code = %[#line #{@line_number + 1} "#{@filename}"#{code}\n#linereset\n]
+          else
+            code = %[#line #{@line_number} "#{@filename}"\n#{code}\n#linereset\n]
+          end
+        end
         if @code_blocks[name]
           @code_blocks[name] += code
         else
@@ -259,6 +331,13 @@ class Propane
       end
     end
 
+    def parse_noline_statement!
+      if md = consume!(/noline\s*;/)
+        @noline = true
+        true
+      end
+    end
+
     def parse_pattern!
       if md = consume!(%r{/})
         pattern = ""
@@ -272,6 +351,8 @@ class Propane
             end
           elsif md = consume!(%r{(.)})
             pattern += md[1]
+          elsif @input == "" || @input.start_with?("\n")
+            raise Error.new("Line #{@line_number}: Unterminated pattern; expected `/`")
           end
         end
         pattern
@@ -280,9 +361,14 @@ class Propane
 
     def parse_code_block!
       if md = consume!(/<<(.*?)>>\n/m)
-        code = md[1]
-        code.sub!(/\A\n/, "")
-        code += "\n" unless code.end_with?("\n")
+        code = md[1].chomp
+        unless @noline
+          if code.start_with?("\n")
+            code = %[#line #{@line_number + 1} "#{@filename}"#{code}\n#linereset\n]
+          else
+            code = %[#line #{@line_number} "#{@filename}"\n#{code}\n#linereset\n]
+          end
+        end
         code
       end
     end

lib/propane/parser.rb

@@ -16,11 +16,20 @@ class Propane
       @warnings = Set.new
       @errors = Set.new
       @options = options
-      start_item = Item.new(grammar.rules.first, 0)
-      eval_item_sets = Set[ItemSet.new([start_item])]
+      start_items = grammar.rules[0...grammar.start_rules.length].map do |start_rule|
+        Item.new(start_rule, 0)
+      end
+      start_item_sets = start_items.map {|item| ItemSet.new([item])}
+      eval_item_sets = Set[*start_item_sets]
 
       while eval_item_sets.size > 0
-        item_set = eval_item_sets.first
+        item_set =
+          if start_item_sets.size > 0
+            # Ensure we evaluate start_item_sets first in order
+            start_item_sets.slice!(0)
+          else
+            eval_item_sets.first
+          end
         eval_item_sets.delete(item_set)
         unless @item_sets_set.include?(item_set)
           item_set.id = @item_sets.size
@@ -64,27 +73,16 @@ class Propane
       @shift_table = []
       @reduce_table = []
       @item_sets.each do |item_set|
-        shift_entries = item_set.next_symbols.map do |next_symbol|
-          state_id =
-            if next_symbol.name == "$EOF"
-              0
-            else
-              item_set.next_item_set[next_symbol].id
-            end
-          {
-            symbol: next_symbol,
-            state_id: state_id,
-          }
-        end
         unless item_set.reduce_rules.empty?
-          shift_entries.each do |shift_entry|
+          item_set.shift_entries.each do |shift_entry|
             token = shift_entry[:symbol]
-            if get_lookahead_reduce_actions_for_item_set(item_set).include?(token)
-              rule = item_set.reduce_actions[token]
+            if item_set.reduce_actions
+              if rule = item_set.reduce_actions[token]
                 @warnings << "Shift/Reduce conflict (state #{item_set.id}) between token #{token.name} and rule #{rule.name} (defined on line #{rule.line_number})"
               end
             end
           end
+        end
         reduce_entries =
           if rule = item_set.reduce_rule
             [{token_id: @grammar.invalid_token_id, rule_id: rule.id, rule: rule,
@@ -92,7 +90,7 @@ class Propane
               propagate_optional_target: rule.optional? && rule.components.size == 1}]
           elsif reduce_actions = item_set.reduce_actions
             reduce_actions.map do |token, rule|
-              {token_id: token.id, rule_id: rule.id, rule: rule,
+              {token: token, token_id: token.id, rule_id: rule.id, rule: rule,
                rule_set_id: rule.rule_set.id, n_states: rule.components.size,
                propagate_optional_target: rule.optional? && rule.components.size == 1}
             end
@@ -101,11 +99,11 @@ class Propane
           end
         @state_table << {
           shift_index: @shift_table.size,
-          n_shifts: shift_entries.size,
+          n_shifts: item_set.shift_entries.size,
           reduce_index: @reduce_table.size,
           n_reduces: reduce_entries.size,
         }
-        @shift_table += shift_entries
+        @shift_table += item_set.shift_entries
         @reduce_table += reduce_entries
       end
     end
@@ -125,8 +123,110 @@ class Propane
     # @return [void]
     def build_reduce_actions!
       @item_sets.each do |item_set|
+        build_shift_entries(item_set)
         build_reduce_actions_for_item_set(item_set)
       end
+      item_sets_to_process = @item_sets.select do |item_set|
+        # We need lookahead reduce actions if:
+        # 1) There is more than one possible rule to reduce. In this case the
+        #    lookahead token can help choose which rule to reduce.
+        # 2) There is at least one shift action and one reduce action for
+        #    this item set. In this case the lookahead reduce actions are
+        #    needed to test for a Shift/Reduce conflict.
+        item_set.reduce_rules.size > 1 ||
+          (item_set.reduce_rules.size > 0 && item_set.shift_entries.size > 0)
+      end
+      if RbConfig::CONFIG["host_os"] =~ /linux/
+        item_sets_by_id = {}
+        item_sets_to_process.each do |item_set|
+          item_sets_by_id[item_set.object_id] = item_set
+        end
+        tokens_by_id = {}
+        @grammar.tokens.each do |token|
+          tokens_by_id[token.object_id] = token
+        end
+        rules_by_id = {}
+        @grammar.rules.each do |rule|
+          rules_by_id[rule.object_id] = rule
+        end
+        n_threads = Util.determine_n_threads
+        semaphore = Mutex.new
+        queue = Queue.new
+        threads = {}
+        n_threads.times do
+          piper, pipew = IO.pipe
+          thread = Thread.new do
+            loop do
+              item_set = nil
+              semaphore.synchronize do
+                item_set = item_sets_to_process.slice!(0)
+              end
+              break if item_set.nil?
+              fork do
+                piper.close
+                build_lookahead_reduce_actions_for_item_set(item_set, pipew)
+              end
+            end
+            queue.push(Thread.current)
+          end
+          threads[thread] = [piper, pipew]
+        end
+        until threads.empty?
+          thread = queue.pop
+          piper, pipew = threads[thread]
+          pipew.close
+          thread_txt = piper.read
+          thread_txt.each_line do |line|
+            if line.start_with?("RA,")
+              parts = line.split(",")
+              item_set_id, token_id, rule_id = parts[1..3].map(&:to_i)
+              item_set = item_sets_by_id[item_set_id]
+              unless item_set
+                raise "Internal error: could not find item set from thread"
+              end
+              token = tokens_by_id[token_id]
+              unless item_set
+                raise "Internal error: could not find token from thread"
+              end
+              rule = rules_by_id[rule_id]
+              unless item_set
+                raise "Internal error: could not find rule from thread"
+              end
+              item_set.reduce_actions ||= {}
+              item_set.reduce_actions[token] = rule
+            elsif line.start_with?("Error: ")
+              @errors << line.chomp
+            else
+              raise "Internal error: unhandled thread line #{line}"
+            end
+          end
+          thread.join
+          threads.delete(thread)
+        end
+      else
+        # Fall back to single threaded algorithm.
+        item_sets_to_process.each do |item_set|
+          item_set.reduce_actions = build_lookahead_reduce_actions_for_item_set(item_set)
+        end
+      end
+    end
+
+    # Build the shift entries for a single item set.
+    #
+    # @return [void]
+    def build_shift_entries(item_set)
+      item_set.shift_entries = item_set.next_symbols.map do |next_symbol|
+        state_id =
+          if next_symbol.name == "$EOF"
+            0
+          else
+            item_set.next_item_set[next_symbol].id
+          end
+        {
+          symbol: next_symbol,
+          state_id: state_id,
+        }
+      end
     end
 
     # Build the reduce actions for a single item set (parser state).
@@ -145,33 +245,18 @@ class Propane
       if item_set.reduce_rules.size == 1
         item_set.reduce_rule = item_set.reduce_rules.first
       end
-
-      if item_set.reduce_rules.size > 1
-        # Force item_set.reduce_actions to be built to store the lookahead
-        # tokens for the possible reduce rules if there is more than one.
-        get_lookahead_reduce_actions_for_item_set(item_set)
-      end
-    end
-
-    # Get the reduce actions for a single item set (parser state).
-    #
-    # @param item_set [ItemSet]
-    #   ItemSet (parser state)
-    #
-    # @return [Hash]
-    #   Mapping of lookahead Tokens to the Rules to reduce.
-    def get_lookahead_reduce_actions_for_item_set(item_set)
-      item_set.reduce_actions ||= build_lookahead_reduce_actions_for_item_set(item_set)
     end
 
     # Build the reduce actions for a single item set (parser state).
     #
     # @param item_set [ItemSet]
     #   ItemSet (parser state)
+    # @param fh [File]
+    #   Output file handle for multiprocessing mode.
     #
     # @return [Hash]
     #   Mapping of lookahead Tokens to the Rules to reduce.
-    def build_lookahead_reduce_actions_for_item_set(item_set)
+    def build_lookahead_reduce_actions_for_item_set(item_set, fh = nil)
       # We will be looking for all possible tokens that can follow instances of
       # these rules. Rather than looking through the entire grammar for the
       # possible following tokens, we will only look in the item sets leading
@@ -182,9 +267,12 @@ class Propane
         lookahead_tokens_for_rule = build_lookahead_tokens_to_reduce(reduce_rule, item_sets)
         lookahead_tokens_for_rule.each do |lookahead_token|
           if existing_reduce_rule = reduce_actions[lookahead_token]
-            @errors << "Error: reduce/reduce conflict (state #{item_set.id}) between rule #{existing_reduce_rule.name}##{existing_reduce_rule.id} (defined on line #{existing_reduce_rule.line_number}) and rule #{reduce_rule.name}##{reduce_rule.id} (defined on line #{reduce_rule.line_number}) for lookahead token #{lookahead_token}"
+            error = "Error: reduce/reduce conflict (state #{item_set.id}) between rule #{existing_reduce_rule.name}##{existing_reduce_rule.id} (defined on line #{existing_reduce_rule.line_number}) and rule #{reduce_rule.name}##{reduce_rule.id} (defined on line #{reduce_rule.line_number}) for lookahead token #{lookahead_token}"
+            @errors << error
+            fh.puts(error) if fh
           end
           reduce_actions[lookahead_token] = reduce_rule
+          fh.puts "RA,#{item_set.object_id},#{lookahead_token.object_id},#{reduce_rule.object_id}" if fh
         end
         reduce_actions
       end
@@ -231,6 +319,7 @@ class Propane
                   rule_set = item.rule.rule_set
                   unless checked_rule_sets.include?(rule_set)
                     rule_sets_to_check_after << rule_set
+                    checked_rule_sets << rule_set
                   end
                   break
                 when Token

lib/propane/parser/item.rb

@@ -22,6 +22,7 @@ class Propane
       def initialize(rule, position)
         @rule = rule
         @position = position
+        @_hash = [@rule, @position].hash
       end
 
       # Hash function.
@@ -29,7 +30,7 @@ class Propane
       # @return [Integer]
       #   Hash code.
       def hash
-        [@rule, @position].hash
+        @_hash
       end
 
       # Compare Item objects.

lib/propane/parser/item_set.rb

@@ -34,6 +34,10 @@ class Propane
       #   more than one rule that could be reduced.
       attr_accessor :reduce_actions
 
+      # @return [Array<Hash>]
+      #   Shift table entries.
+      attr_accessor :shift_entries
+
       # Build an ItemSet.
       #
       # @param items [Array<Item>]

lib/propane/pattern.rb

@@ -55,7 +55,7 @@ class Propane
       @line_number = options[:line_number]
       @modes = options[:modes]
       @ptypename = options[:ptypename]
-      regex = Regex.new(@pattern)
+      regex = Regex.new(@pattern, @line_number)
       regex.nfa.end_state.accepts = self
       @nfa = regex.nfa
     end

lib/propane/regex.rb

@@ -4,12 +4,13 @@ class Propane
     attr_reader :unit
     attr_reader :nfa
 
-    def initialize(pattern)
+    def initialize(pattern, line_number)
       @pattern = pattern.dup
+      @line_number = line_number
       @unit = parse_alternates
       @nfa = @unit.to_nfa
       if @pattern != ""
-        raise Error.new(%[Unexpected "#{@pattern}" in pattern])
+        raise Error.new(%[Line #{@line_number}: unexpected "#{@pattern}" in pattern])
       end
     end
 
@@ -41,7 +42,7 @@ class Propane
             mu = MultiplicityUnit.new(last_unit, min_count, max_count)
             au.replace_last!(mu)
           else
-            raise Error.new("#{c} follows nothing")
+            raise Error.new("Line #{@line_number}: #{c} follows nothing")
           end
         when "|"
           au.new_alternate!
@@ -59,7 +60,7 @@ class Propane
     def parse_group
       au = parse_alternates
       if @pattern[0] != ")"
-        raise Error.new("Unterminated group in pattern")
+        raise Error.new("Line #{@line_number}: unterminated group in pattern")
       end
       @pattern.slice!(0)
       au
@@ -70,7 +71,7 @@ class Propane
       index = 0
       loop do
         if @pattern == ""
-          raise Error.new("Unterminated character class")
+          raise Error.new("Line #{@line_number}: unterminated character class")
         end
         c = @pattern.slice!(0)
         if c == "]"
@@ -84,13 +85,13 @@ class Propane
         elsif c == "-" && @pattern[0] != "]"
           begin_cu = ccu.last_unit
           unless begin_cu.is_a?(CharacterRangeUnit) && begin_cu.code_point_range.size == 1
-            raise Error.new("Character range must be between single characters")
+            raise Error.new("Line #{@line_number}: character range must be between single characters")
           end
           if @pattern[0] == "\\"
             @pattern.slice!(0)
             end_cu = parse_backslash
             unless end_cu.is_a?(CharacterRangeUnit) && end_cu.code_point_range.size == 1
-              raise Error.new("Character range must be between single characters")
+              raise Error.new("Line #{@line_number}: character range must be between single characters")
             end
             max_code_point = end_cu.code_point
           else
@@ -116,7 +117,7 @@ class Propane
         elsif max_count.to_s != ""
           max_count = max_count.to_i
           if max_count < min_count
-            raise Error.new("Maximum repetition count cannot be less than minimum repetition count")
+            raise Error.new("Line #{@line_number}: maximum repetition count cannot be less than minimum repetition count")
           end
         else
           max_count = nil
@@ -124,28 +125,33 @@ class Propane
         @pattern = pattern
         [min_count, max_count]
       else
-        raise Error.new("Unexpected match count at #{@pattern}")
+        raise Error.new("Line #{@line_number}: unexpected match count following {")
       end
     end
 
     def parse_backslash
       if @pattern == ""
-        raise Error.new("Error: unfollowed \\")
+        raise Error.new("Line #{@line_number}: error: unfollowed \\")
       else
         c = @pattern.slice!(0)
         case c
         when "a"
-          CharacterRangeUnit.new("\a", "\a")
+          CharacterRangeUnit.new("\a")
         when "b"
-          CharacterRangeUnit.new("\b", "\b")
+          CharacterRangeUnit.new("\b")
         when "d"
           CharacterRangeUnit.new("0", "9")
+        when "D"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu.negate = true
+          ccu
         when "f"
-          CharacterRangeUnit.new("\f", "\f")
+          CharacterRangeUnit.new("\f")
         when "n"
-          CharacterRangeUnit.new("\n", "\n")
+          CharacterRangeUnit.new("\n")
         when "r"
-          CharacterRangeUnit.new("\r", "\r")
+          CharacterRangeUnit.new("\r")
         when "s"
           ccu = CharacterClassUnit.new
           ccu << CharacterRangeUnit.new(" ")
@@ -155,10 +161,35 @@ class Propane
           ccu << CharacterRangeUnit.new("\f")
           ccu << CharacterRangeUnit.new("\v")
           ccu
+        when "S"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new(" ")
+          ccu << CharacterRangeUnit.new("\t")
+          ccu << CharacterRangeUnit.new("\r")
+          ccu << CharacterRangeUnit.new("\n")
+          ccu << CharacterRangeUnit.new("\f")
+          ccu << CharacterRangeUnit.new("\v")
+          ccu.negate = true
+          ccu
         when "t"
-          CharacterRangeUnit.new("\t", "\t")
+          CharacterRangeUnit.new("\t")
         when "v"
-          CharacterRangeUnit.new("\v", "\v")
+          CharacterRangeUnit.new("\v")
+        when "w"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("_")
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu << CharacterRangeUnit.new("a", "z")
+          ccu << CharacterRangeUnit.new("A", "Z")
+          ccu
+        when "W"
+          ccu = CharacterClassUnit.new
+          ccu << CharacterRangeUnit.new("_")
+          ccu << CharacterRangeUnit.new("0", "9")
+          ccu << CharacterRangeUnit.new("a", "z")
+          ccu << CharacterRangeUnit.new("A", "Z")
+          ccu.negate = true
+          ccu
         else
           CharacterRangeUnit.new(c)
         end

lib/propane/regex/unit.rb

@@ -92,17 +92,20 @@ class Propane
         @units = []
         @negate = false
       end
-      def initialize
-        @units = []
-      end
-      def method_missing(*args)
-        @units.__send__(*args)
+      def method_missing(*args, &block)
+        @units.__send__(*args, &block)
       end
       def <<(thing)
         if thing.is_a?(CharacterClassUnit)
+          if thing.negate
+            CodePointRange.invert_ranges(thing.map(&:code_point_range)).each do |cpr|
+              CharacterRangeUnit.new(cpr.first, cpr.last)
+            end
+          else
             thing.each do |ccu_unit|
               @units << ccu_unit
             end
+          end
         else
           @units << thing
         end

lib/propane/rule.rb

@@ -36,7 +36,7 @@ class Propane
 
     # @return [Array<Integer>]
     #   Map this rule's components to their positions in the parent RuleSet's
-    #   node field pointer array. This is used for AST construction.
+    #   node field pointer array. This is used for tree construction.
     attr_accessor :rule_set_node_field_index_map
 
     # Construct a Rule.

lib/propane/rule_set.rb

@@ -4,8 +4,8 @@ class Propane
   class RuleSet
 
     # @return [Array<Hash>]
-    #   AST fields.
-    attr_reader :ast_fields
+    #   tree fields.
+    attr_reader :tree_fields
 
     # @return [Integer]
     #   ID of the RuleSet.
@@ -100,28 +100,28 @@ class Propane
 
     # Finalize a RuleSet after adding all Rules to it.
     def finalize(grammar)
-      if grammar.ast
-        build_ast_fields(grammar)
+      if grammar.tree
+        build_tree_fields(grammar)
       end
     end
 
     private
 
-    # Build the set of AST fields for this RuleSet.
+    # Build the set of tree fields for this RuleSet.
     #
     # This is an Array of Hashes. Each entry in the Array corresponds to a
-    # field location in the AST node. The entry is a Hash. It could have one or
+    # field location in the tree node. The entry is a Hash. It could have one or
     # two keys. It will always have the field name with a positional suffix as
     # a key. It may also have the field name without the positional suffix if
     # that field only exists in one position across all Rules in the RuleSet.
     #
     # @return [void]
-    def build_ast_fields(grammar)
-      field_ast_node_indexes = {}
+    def build_tree_fields(grammar)
+      field_tree_node_indexes = {}
       field_indexes_across_all_rules = {}
-      # Stores the index into @ast_fields by field alias name.
+      # Stores the index into @tree_fields by field alias name.
       field_aliases = {}
-      @ast_fields = []
+      @tree_fields = []
       @rules.each do |rule|
         rule.components.each_with_index do |component, i|
           if component.is_a?(RuleSet) && component.optional?
@@ -132,25 +132,25 @@ class Propane
           else
             node_name = component.name
           end
-          struct_name = "#{grammar.ast_prefix}#{node_name}#{grammar.ast_suffix}"
+          struct_name = "#{grammar.tree_prefix}#{node_name}#{grammar.tree_suffix}"
           field_name = "p#{node_name}#{i + 1}"
-          unless field_ast_node_indexes[field_name]
-            field_ast_node_indexes[field_name] = @ast_fields.size
-            @ast_fields << {field_name => struct_name}
+          unless field_tree_node_indexes[field_name]
+            field_tree_node_indexes[field_name] = @tree_fields.size
+            @tree_fields << {field_name => struct_name}
           end
           rule.aliases.each do |alias_name, index|
             if index == i
-              alias_ast_fields_index = field_ast_node_indexes[field_name]
-              if field_aliases[alias_name] && field_aliases[alias_name] != alias_ast_fields_index
-                raise Error.new("Error: conflicting AST node field positions for alias `#{alias_name}` in rule #{rule.name} defined on line #{rule.line_number}")
+              alias_tree_fields_index = field_tree_node_indexes[field_name]
+              if field_aliases[alias_name] && field_aliases[alias_name] != alias_tree_fields_index
+                raise Error.new("Error: conflicting tree node field positions for alias `#{alias_name}` in rule #{rule.name} defined on line #{rule.line_number}")
               end
-              field_aliases[alias_name] = alias_ast_fields_index
-              @ast_fields[alias_ast_fields_index][alias_name] = @ast_fields[alias_ast_fields_index].first[1]
+              field_aliases[alias_name] = alias_tree_fields_index
+              @tree_fields[alias_tree_fields_index][alias_name] = @tree_fields[alias_tree_fields_index].first[1]
             end
           end
           field_indexes_across_all_rules[node_name] ||= Set.new
-          field_indexes_across_all_rules[node_name] << field_ast_node_indexes[field_name]
-          rule.rule_set_node_field_index_map[i] = field_ast_node_indexes[field_name]
+          field_indexes_across_all_rules[node_name] << field_tree_node_indexes[field_name]
+          rule.rule_set_node_field_index_map[i] = field_tree_node_indexes[field_name]
         end
       end
       field_indexes_across_all_rules.each do |node_name, indexes_across_all_rules|
@@ -158,8 +158,8 @@ class Propane
           # If this field was only seen in one position across all rules,
           # then add an alias to the positional field name that does not
           # include the position.
-          @ast_fields[indexes_across_all_rules.first]["p#{node_name}"] =
-            "#{grammar.ast_prefix}#{node_name}#{grammar.ast_suffix}"
+          @tree_fields[indexes_across_all_rules.first]["p#{node_name}"] =
+            "#{grammar.tree_prefix}#{node_name}#{grammar.tree_suffix}"
         end
       end
     end

lib/propane/util.rb

@@ -10,6 +10,32 @@ class Propane
         "#{s}\n* #{message} *\n#{s}\n"
       end
 
+      # Determine the number of threads to use.
+      #
+      # @return [Integer]
+      #   The number of threads to use.
+      def determine_n_threads
+        # Try to figure out how many threads are available on the host hardware.
+        begin
+          case RbConfig::CONFIG["host_os"]
+          when /linux/
+            return File.read("/proc/cpuinfo").scan(/^processor\s*:/).size
+          when /mswin|mingw|msys/
+            if `wmic cpu get NumberOfLogicalProcessors -value` =~ /NumberOfLogicalProcessors=(\d+)/
+              return $1.to_i
+            end
+          when /darwin/
+            if `sysctl -n hw.ncpu` =~ /(\d+)/
+              return $1.to_i
+            end
+          end
+        rescue
+        end
+
+        # If we can't figure it out, default to 4.
+        4
+      end
+
     end
 
   end

lib/propane/version.rb

@@ -1,3 +1,3 @@
 class Propane
-  VERSION = "2.2.0"
+  VERSION = "4.7.0"
 end

spec/json_parser.c.propane

@@ -120,11 +120,11 @@ string: /\\t/ <<
 >>
 string: /\\u[0-9a-fA-F]{4}/ <<
   /* Not actually going to encode the code point for this example... */
-  char s[] = {'{', match[2], match[3], match[4], match[5], '}', 0};
+  char s[] = {'{', (char)match[2], (char)match[3], (char)match[4], (char)match[5], '}', 0};
   str_append(&string_value, s);
 >>
 string: /[^\\]/ <<
-  char s[] = {match[0], 0};
+  char s[] = {(char)match[0], 0};
   str_append(&string_value, s);
 >>
 Start -> Value <<

spec/json_types.c

@@ -5,7 +5,7 @@
 
 JSONValue * JSONValue_new(size_t id)
 {
-    JSONValue * jv = calloc(1, sizeof(JSONValue));
+    JSONValue * jv = (JSONValue *)calloc(1, sizeof(JSONValue));
     jv->id = id;
     return jv;
 }
@@ -29,7 +29,7 @@ void JSONObject_append(JSONValue * object, char const * name, JSONValue * value)
         }
     }
     size_t const new_size = size + 1;
-    void * new_entries = malloc(sizeof(object->object.entries[0]) * new_size);
+    JSONObjectEntry * new_entries = (JSONObjectEntry *)malloc(sizeof(object->object.entries[0]) * new_size);
     if (size > 0)
     {
         memcpy(new_entries, object->object.entries, size * sizeof(object->object.entries[0]));
@@ -52,7 +52,7 @@ void JSONArray_append(JSONValue * array, JSONValue * value)
 {
     size_t const size = array->array.size;
     size_t const new_size = size + 1;
-    JSONValue ** new_entries = malloc(sizeof(JSONValue *) * new_size);
+    JSONValue ** new_entries = (JSONValue **)malloc(sizeof(JSONValue *) * new_size);
     if (array->array.size > 0)
     {
         memcpy(new_entries, array->array.entries, sizeof(JSONValue *) * size);

spec/json_types.h

@@ -11,6 +11,12 @@
 #define JSON_FALSE 5u
 #define JSON_NULL 6u
 
+typedef struct JSONObjectEntry_s
+{
+    char const * name;
+    struct JSONValue_s * value;
+} JSONObjectEntry;
+
 typedef struct JSONValue_s
 {
     size_t id;
@@ -19,11 +25,7 @@ typedef struct JSONValue_s
         struct
         {
             size_t size;
-            struct
-            {
-                char const * name;
-                struct JSONValue_s * value;
-            } * entries;
+            JSONObjectEntry * entries;
         } object;
         struct
         {

spec/propane/grammar_spec.rb

@@ -28,7 +28,7 @@ B -> <<
   b = 0;
 >>
 EOF
-      grammar = Grammar.new(input)
+      grammar = Grammar.new(input, "test.propane")
       expect(grammar.modulename).to eq "a.b"
       expect(grammar.ptype).to eq "XYZ  *"
       expect(grammar.ptypes).to eq("default" => "XYZ  *")
@@ -62,7 +62,7 @@ EOF
       expect(o).to_not be_nil
       expect(o.pattern).to eq "token_with_code"
       expect(o.line_number).to eq 11
-      expect(o.code).to eq "Code for the token\n"
+      expect(o.code).to eq %[#line 12 "test.propane"\nCode for the token\n#linereset\n]
 
       o = grammar.tokens.find {|token| token.name == "token_with_no_pattern"}
       expect(o).to_not be_nil
@@ -83,7 +83,7 @@ EOF
       expect(o.name).to eq "A"
       expect(o.components).to eq %w[B]
       expect(o.line_number).to eq 19
-      expect(o.code).to eq "  a = 42;\n"
+      expect(o.code).to eq %[#line 20 "test.propane"\n  a = 42;\n#linereset\n]
 
       o = grammar.rules[1]
       expect(o.name).to eq "B"
@@ -95,7 +95,7 @@ EOF
       expect(o.name).to eq "B"
       expect(o.components).to eq []
       expect(o.line_number).to eq 23
-      expect(o.code).to eq "  b = 0;\n"
+      expect(o.code).to eq %[#line 24 "test.propane"\n  b = 0;\n#linereset\n]
     end
 
     it "parses code segments with semicolons" do
@@ -113,7 +113,7 @@ tokenid token_with_no_pattern;
 
 prefix myparser_;
 EOF
-      grammar = Grammar.new(input)
+      grammar = Grammar.new(input, "test.propane")
       expect(grammar.prefix).to eq "myparser_"
 
       o = grammar.tokens.find {|token| token.name == "code1"}
@@ -122,7 +122,7 @@ EOF
 
       o = grammar.patterns.find {|pattern| pattern.token == o}
       expect(o).to_not be_nil
-      expect(o.code).to eq "  a = b;\n  return c;\n"
+      expect(o.code).to eq %[#line 2 "test.propane"\n  a = b;\n  return c;\n#linereset\n]
 
       o = grammar.tokens.find {|token| token.name == "code2"}
       expect(o).to_not be_nil
@@ -130,7 +130,42 @@ EOF
 
       o = grammar.patterns.find {|pattern| pattern.token == o}
       expect(o).to_not be_nil
-      expect(o.code).to eq %[  writeln("Hello there");\n]
+      expect(o.code).to eq %[#line 7 "test.propane"\n  writeln("Hello there");\n#linereset\n]
+    end
+
+    it "does not emit #line directives with noline statement" do
+      input = <<EOF
+noline;
+
+token code1 <<
+  a = b;
+  return c;
+>>
+
+token code2 <<
+  writeln("Hello there");
+>>
+
+tokenid token_with_no_pattern;
+
+prefix myparser_;
+EOF
+      grammar = Grammar.new(input, "test.propane")
+      expect(grammar.prefix).to eq "myparser_"
+
+      o = grammar.tokens.find {|token| token.name == "code1"}
+      expect(o).to_not be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.code).to eq %[\n  a = b;\n  return c;]
+
+      o = grammar.tokens.find {|token| token.name == "code2"}
+      expect(o).to_not be_nil
+
+      o = grammar.patterns.find {|pattern| pattern.token == o}
+      expect(o).to_not be_nil
+      expect(o.code).to eq %[\n  writeln("Hello there");]
     end
 
     it "supports mode labels" do
@@ -144,7 +179,7 @@ m2: /bar/ <<
 drop /q/;
 m3: drop /r/;
 EOF
-      grammar = Grammar.new(input)
+      grammar = Grammar.new(input, "test.propane")
 
       o = grammar.tokens.find {|token| token.name == "a"}
       expect(o).to_not be_nil
@@ -197,7 +232,7 @@ tokenid int(integer);
 Start (node) -> R;
 R -> abc int;
 EOF
-      grammar = Grammar.new(input)
+      grammar = Grammar.new(input, "test.propane")
 
       o = grammar.tokens.find {|token| token.name == "abc"}
       expect(o).to_not be_nil

spec/propane/lexer/dfa_spec.rb

@@ -51,7 +51,7 @@ class TestLexer
 end
 
 def run(grammar, input)
-  grammar = Propane::Grammar.new(grammar)
+  grammar = Propane::Grammar.new(grammar, "test.propane")
   token_dfa = Propane::Lexer::DFA.new(grammar.patterns)
   test_lexer = TestLexer.new(token_dfa)
   test_lexer.lex(input)
@@ -126,6 +126,74 @@ EOF
     ]
     expect(run(<<EOF, ";")).to eq expected
 token semicolon /;/;
+EOF
+  end
+
+  it "matches a negated character class" do
+    expected = [
+      ["pattern", "/abc/"],
+    ]
+    expect(run(<<EOF, "/abc/")).to eq expected
+token pattern /\\/[^\\s]*\\//;
+EOF
+  end
+
+  it "matches special character classes " do
+    expected = [
+      ["a", "abc123_FOO"],
+    ]
+    expect(run(<<EOF, "abc123_FOO")).to eq expected
+token a /\\w+/;
+EOF
+    expected = [
+      ["b", "FROG*%$#"],
+    ]
+    expect(run(<<EOF, "FROG*%$#")).to eq expected
+token b /FROG\\D{1,4}/;
+EOF
+    expected = [
+      ["c", "$883366"],
+    ]
+    expect(run(<<EOF, "$883366")).to eq expected
+token c /$\\d+/;
+EOF
+    expected = [
+      ["d", "^&$@"],
+    ]
+    expect(run(<<EOF, "^&$@")).to eq expected
+token d /^\\W+/;
+EOF
+    expected = [
+      ["a", "abc123_FOO"],
+      [nil, " "],
+      ["b", "FROG*%$#"],
+      [nil, " "],
+      ["c", "$883366"],
+      [nil, " "],
+      ["d", "^&$@"],
+    ]
+    expect(run(<<EOF, "abc123_FOO FROG*%$# $883366 ^&$@")).to eq expected
+token a /\\w+/;
+token b /FROG\\D{1,4}/;
+token c /$\\d+/;
+token d /^\\W+/;
+drop /\\s+/;
+EOF
+  end
+
+  it "matches a negated character class with a nested inner negated character class" do
+    expected = [
+      ["t", "$&*"],
+    ]
+    expect(run(<<EOF, "$&*")).to eq expected
+token t /[^%\\W]+/;
+EOF
+  end
+
+  it "\\s matches a newline" do
+    expected = [["s", "\n"]]
+    expect(run(<<EOF, "\n")).to eq expected
+token s /\\s/;
 EOF
   end
 end

spec/propane/regex_spec.rb

@@ -2,14 +2,14 @@ class Propane
   RSpec.describe Regex do
 
     it "parses an empty expression" do
-      regex = Regex.new("")
+      regex = Regex.new("", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0].size).to eq 0
     end
 
     it "parses a single character unit expression" do
-      regex = Regex.new("a")
+      regex = Regex.new("a", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -19,7 +19,7 @@ class Propane
     end
 
     it "parses a group with a single character unit expression" do
-      regex = Regex.new("(a)")
+      regex = Regex.new("(a)", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -33,7 +33,7 @@ class Propane
     end
 
     it "parses a *" do
-      regex = Regex.new("a*")
+      regex = Regex.new("a*", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -47,7 +47,7 @@ class Propane
     end
 
     it "parses a +" do
-      regex = Regex.new("a+")
+      regex = Regex.new("a+", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -61,7 +61,7 @@ class Propane
     end
 
     it "parses a ?" do
-      regex = Regex.new("a?")
+      regex = Regex.new("a?", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -75,7 +75,7 @@ class Propane
     end
 
     it "parses a multiplicity count" do
-      regex = Regex.new("a{5}")
+      regex = Regex.new("a{5}", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -89,7 +89,7 @@ class Propane
     end
 
     it "parses a minimum-only multiplicity count" do
-      regex = Regex.new("a{5,}")
+      regex = Regex.new("a{5,}", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -103,7 +103,7 @@ class Propane
     end
 
     it "parses a minimum and maximum multiplicity count" do
-      regex = Regex.new("a{5,8}")
+      regex = Regex.new("a{5,8}", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -118,7 +118,7 @@ class Propane
     end
 
     it "parses an escaped *" do
-      regex = Regex.new("a\\*")
+      regex = Regex.new("a\\*", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -131,7 +131,7 @@ class Propane
     end
 
     it "parses an escaped +" do
-      regex = Regex.new("a\\+")
+      regex = Regex.new("a\\+", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -144,7 +144,7 @@ class Propane
     end
 
     it "parses an escaped \\" do
-      regex = Regex.new("\\\\d")
+      regex = Regex.new("\\\\d", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -157,7 +157,7 @@ class Propane
     end
 
     it "parses a character class" do
-      regex = Regex.new("[a-z_]")
+      regex = Regex.new("[a-z_]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -175,7 +175,7 @@ class Propane
     end
 
     it "parses a negated character class" do
-      regex = Regex.new("[^xyz]")
+      regex = Regex.new("[^xyz]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -189,8 +189,25 @@ class Propane
       expect(ccu[0].first).to eq "x".ord
     end
 
+    it "parses a negated character class with inner character classes" do
+      regex = Regex.new("[^x\\sz]", 1)
+      expect(regex.unit).to be_a Regex::AlternatesUnit
+      expect(regex.unit.alternates.size).to eq 1
+      expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
+      seq_unit = regex.unit.alternates[0]
+      expect(seq_unit.size).to eq 1
+      expect(seq_unit[0]).to be_a Regex::CharacterClassUnit
+      ccu = seq_unit[0]
+      expect(ccu.negate).to be_truthy
+      expect(ccu.size).to eq 8
+      expect(ccu[0]).to be_a Regex::CharacterRangeUnit
+      expect(ccu[0].first).to eq "x".ord
+      expect(ccu[1].first).to eq " ".ord
+      expect(ccu[7].first).to eq "z".ord
+    end
+
     it "parses - as a plain character at beginning of a character class" do
-      regex = Regex.new("[-9]")
+      regex = Regex.new("[-9]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -204,7 +221,7 @@ class Propane
     end
 
     it "parses - as a plain character at end of a character class" do
-      regex = Regex.new("[0-]")
+      regex = Regex.new("[0-]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -220,7 +237,7 @@ class Propane
     end
 
     it "parses - as a plain character at beginning of a negated character class" do
-      regex = Regex.new("[^-9]")
+      regex = Regex.new("[^-9]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -235,7 +252,7 @@ class Propane
     end
 
     it "parses . as a plain character in a character class" do
-      regex = Regex.new("[.]")
+      regex = Regex.new("[.]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -250,7 +267,7 @@ class Propane
     end
 
     it "parses - as a plain character when escaped in middle of character class" do
-      regex = Regex.new("[0\\-9]")
+      regex = Regex.new("[0\\-9]", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -269,7 +286,7 @@ class Propane
     end
 
     it "parses alternates" do
-      regex = Regex.new("ab|c")
+      regex = Regex.new("ab|c", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 2
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -279,7 +296,7 @@ class Propane
     end
 
     it "parses a ." do
-      regex = Regex.new("a.b")
+      regex = Regex.new("a.b", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 1
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit
@@ -290,7 +307,7 @@ class Propane
     end
 
     it "parses something complex" do
-      regex = Regex.new("(a|)*|[^^]|\\|v|[x-y]+")
+      regex = Regex.new("(a|)*|[^^]|\\|v|[x-y]+", 1)
       expect(regex.unit).to be_a Regex::AlternatesUnit
       expect(regex.unit.alternates.size).to eq 4
       expect(regex.unit.alternates[0]).to be_a Regex::SequenceUnit

spec/spec_helper.rb

@@ -2,6 +2,10 @@ unless ENV["dist_specs"]
   require "bundler/setup"
   require "simplecov"
 
+  class MyFormatter
+    def format(*args)
+    end
+  end
   SimpleCov.start do
     add_filter "/spec/"
     add_filter "/.bundle/"
@@ -12,6 +16,7 @@ unless ENV["dist_specs"]
     end
     project_name "Propane"
     merge_timeout 3600
+    formatter(MyFormatter)
   end
 
   RSpec.configure do |config|

spec/test_basic_math_grammar.c

@@ -5,25 +5,29 @@
 int main()
 {
     char const * input = "1 + 2 * 3 + 4";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(11, p_result(&context));
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(11, p_result(context));
+    p_context_delete(context);
 
     input = "1 * 2 ** 4 * 3";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(48, p_result(&context));
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(48, p_result(context));
+    p_context_delete(context);
 
     input = "(1 + 2) * 3 + 4";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(13, p_result(&context));
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(13, p_result(context));
+    p_context_delete(context);
 
     input = "(2 * 2) ** 3 + 4 + 5";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(73, p_result(&context));
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(73, p_result(context));
+    p_context_delete(context);
 
     return 0;
 }

spec/test_basic_math_grammar.d

@@ -10,23 +10,23 @@ int main()
 unittest
 {
     string input = "1 + 2 * 3 + 4";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(11, p_result(&context));
+    p_context_t * context;
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(11, p_result(context));
 
     input = "1 * 2 ** 4 * 3";
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(48, p_result(&context));
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(48, p_result(context));
 
     input = "(1 + 2) * 3 + 4";
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(13, p_result(&context));
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(13, p_result(context));
 
     input = "(2 * 2) ** 3 + 4 + 5";
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    assert_eq(73, p_result(&context));
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    assert_eq(73, p_result(context));
 }

spec/test_custom_lex_fn.c

@@ -0,0 +1,15 @@
+#include "testparser.h"
+#include "testutils.h"
+#include <string.h>
+
+int main()
+{
+    char const * input = "cbacba";
+    p_context_t * context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    size_t result = p_result(context);
+    assert_eq(0x932187932187, result);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_custom_lex_fn.d

@@ -0,0 +1,18 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "cbacba";
+    p_context_t * context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    size_t result = p_result(context);
+    assert_eq(0x932187932187, result);
+    p_context_delete(context);
+}

spec/test_drop_code_block.c

@@ -0,0 +1,15 @@
+#include "testparser.h"
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "    # comment 1\n#    comment 2\na\n";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_drop_code_block.d

@@ -0,0 +1,16 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "    # comment 1\n#    comment 2\na\n";
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+}

spec/test_error_positions.c

@@ -5,38 +5,43 @@
 int main()
 {
     char const * input = "a 42";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "a\n123\na  a";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context).row == 3);
-    assert(p_position(&context).col == 4);
-    assert(p_token(&context) == TOKEN_a);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context).row == 3);
+    assert(p_position(context).col == 4);
+    assert(p_token(context) == TOKEN_a);
+    p_context_delete(context);
 
     input = "12";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context).row == 1);
-    assert(p_position(&context).col == 1);
-    assert(p_token(&context) == TOKEN_num);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context).row == 1);
+    assert(p_position(context).col == 1);
+    assert(p_token(context) == TOKEN_num);
+    p_context_delete(context);
 
     input = "a 12\n\nab";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
-    assert(p_position(&context).row == 3);
-    assert(p_position(&context).col == 2);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_UNEXPECTED_INPUT);
+    assert(p_position(context).row == 3);
+    assert(p_position(context).col == 2);
+    p_context_delete(context);
 
     input = "a 12\n\na\n\n77\na   \xAA";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_DECODE_ERROR);
-    assert(p_position(&context).row == 6);
-    assert(p_position(&context).col == 5);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_DECODE_ERROR);
+    assert(p_position(context).row == 6);
+    assert(p_position(context).col == 5);
 
     assert(strcmp(p_token_names[TOKEN_a], "a") == 0);
     assert(strcmp(p_token_names[TOKEN_num], "num") == 0);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_error_positions.d

@@ -9,31 +9,31 @@ int main()
 unittest
 {
     string input = "a 42";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = "a\n123\na  a";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context) == p_position_t(3, 4));
-    assert(p_token(&context) == TOKEN_a);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context) == p_position_t(3, 4));
+    assert(p_token(context) == TOKEN_a);
 
     input = "12";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context) == p_position_t(1, 1));
-    assert(p_token(&context) == TOKEN_num);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context) == p_position_t(1, 1));
+    assert(p_token(context) == TOKEN_num);
 
     input = "a 12\n\nab";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
-    assert(p_position(&context) == p_position_t(3, 2));
+    context = p_context_new(input);
+    assert(p_parse(context) == P_UNEXPECTED_INPUT);
+    assert(p_position(context) == p_position_t(3, 2));
 
     input = "a 12\n\na\n\n77\na   \xAA";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_DECODE_ERROR);
-    assert(p_position(&context) == p_position_t(6, 5));
+    context = p_context_new(input);
+    assert(p_parse(context) == P_DECODE_ERROR);
+    assert(p_position(context) == p_position_t(6, 5));
 
     assert(p_token_names[TOKEN_a] == "a");
     assert(p_token_names[TOKEN_num] == "num");

spec/test_field_aliases.c

@@ -6,8 +6,9 @@
 int main()
 {
     char const * input = "foo1\nbar2";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
     return 0;
 }

spec/test_field_aliases.d

@@ -9,7 +9,7 @@ int main()
 unittest
 {
     string input = "foo1\nbar2";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_lexer.c

@@ -38,73 +38,75 @@ int main()
 
     p_token_info_t token_info;
     char const * input = "5 + 4 * \n677 + 567";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 1u);
     assert(token_info.position.col == 1u);
     assert(token_info.end_position.row == 1u);
     assert(token_info.end_position.col == 1u);
     assert(token_info.length == 1u);
     assert(token_info.token == TOKEN_int);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 1u);
     assert(token_info.position.col == 3u);
     assert(token_info.end_position.row == 1u);
     assert(token_info.end_position.col == 3u);
     assert(token_info.length == 1u);
     assert(token_info.token == TOKEN_plus);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 1u);
     assert(token_info.position.col == 5u);
     assert(token_info.end_position.row == 1u);
     assert(token_info.end_position.col == 5u);
     assert(token_info.length == 1u);
     assert(token_info.token == TOKEN_int);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 1u);
     assert(token_info.position.col == 7u);
     assert(token_info.end_position.row == 1u);
     assert(token_info.end_position.col == 7u);
     assert(token_info.length == 1u);
     assert(token_info.token == TOKEN_times);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 2u);
     assert(token_info.position.col == 1u);
     assert(token_info.end_position.row == 2u);
     assert(token_info.end_position.col == 3u);
     assert(token_info.length == 3u);
     assert(token_info.token == TOKEN_int);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 2u);
     assert(token_info.position.col == 5u);
     assert(token_info.end_position.row == 2u);
     assert(token_info.end_position.col == 5u);
     assert(token_info.length == 1u);
     assert(token_info.token == TOKEN_plus);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 2u);
     assert(token_info.position.col == 7u);
     assert(token_info.end_position.row == 2u);
     assert(token_info.end_position.col == 9u);
     assert(token_info.length == 3u);
     assert(token_info.token == TOKEN_int);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 2u);
     assert(token_info.position.col == 10u);
     assert(token_info.end_position.row == 2u);
     assert(token_info.end_position.col == 10u);
     assert(token_info.length == 0u);
     assert(token_info.token == TOKEN___EOF);
+    p_context_delete(context);
 
-    p_context_init(&context, (uint8_t const *)"", 0u);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)"", 0u);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info.position.row == 1u);
     assert(token_info.position.col == 1u);
     assert(token_info.end_position.row == 1u);
     assert(token_info.end_position.col == 1u);
     assert(token_info.length == 0u);
     assert(token_info.token == TOKEN___EOF);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer.d

@@ -44,26 +44,26 @@ unittest
 {
     p_token_info_t token_info;
     string input = "5 + 4 * \n677 + 567";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(1, 1), p_position_t(1, 1), 1, TOKEN_int));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(1, 3), p_position_t(1, 3), 1, TOKEN_plus));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(1, 5), p_position_t(1, 5), 1, TOKEN_int));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(1, 7), p_position_t(1, 7), 1, TOKEN_times));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(2, 1), p_position_t(2, 3), 3, TOKEN_int));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(2, 5), p_position_t(2, 5), 1, TOKEN_plus));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(2, 7), p_position_t(2, 9), 3, TOKEN_int));
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(2, 10), p_position_t(2, 10), 0, TOKEN___EOF));
 
-    p_context_init(&context, "");
-    assert(p_lex(&context, &token_info) == P_SUCCESS);
+    context = p_context_new("");
+    assert(p_lex(context, &token_info) == P_SUCCESS);
     assert(token_info == p_token_info_t(p_position_t(1, 1), p_position_t(1, 1), 0, TOKEN___EOF));
 }

spec/test_lexer_match_text.c

@@ -6,10 +6,11 @@
 int main()
 {
     char const * input = "identifier_123";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass1\n");
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer_match_text.d

@@ -9,8 +9,8 @@ int main()
 unittest
 {
     string input = `identifier_123`;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass1");
 }

spec/test_lexer_modes.c

@@ -6,15 +6,17 @@
 int main()
 {
     char const * input = "abc \"a string\" def";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass1\n");
+    p_context_delete(context);
 
     input = "abc \"abc def\" def";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass2\n");
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer_modes.d

@@ -9,13 +9,13 @@ int main()
 unittest
 {
     string input = `abc "a string" def`;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass1");
 
     input = `abc "abc def" def`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass2");
 }

spec/test_lexer_multiple_modes.c

@@ -6,15 +6,17 @@
 int main()
 {
     char const * input = "abc.def";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass1\n");
+    p_context_delete(context);
 
     input = "abc .  abc";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass2\n");
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer_multiple_modes.d

@@ -9,13 +9,13 @@ int main()
 unittest
 {
     string input = `abc.def`;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass1");
 
     input = `abc .  abc`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass2");
 }

spec/test_lexer_positions.c

@@ -0,0 +1,50 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+
+int main()
+{
+    char const * input = "abc\n  defg hi\n!";
+    p_context_t * context = p_context_new((uint8_t const *)input, strlen(input));
+    p_token_info_t token_info;
+
+    /* First token "abc" on row 1, cols 1-3. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context->last_start.row == 1u);
+    assert(context->last_start.col == 1u);
+    assert(context->last_end.row == 1u);
+    assert(context->last_end.col == 3u);
+    /* The lexer code block observed the same positions reported to the caller. */
+    assert(context->last_start.row == token_info.position.row);
+    assert(context->last_start.col == token_info.position.col);
+    assert(context->last_end.row == token_info.end_position.row);
+    assert(context->last_end.col == token_info.end_position.col);
+
+    /* Second token "defg" on row 2, cols 3-6. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context->last_start.row == 2u);
+    assert(context->last_start.col == 3u);
+    assert(context->last_end.row == 2u);
+    assert(context->last_end.col == 6u);
+
+    /* Third token "hi" on row 2, cols 8-9. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context->last_start.row == 2u);
+    assert(context->last_start.col == 8u);
+    assert(context->last_end.row == 2u);
+    assert(context->last_end.col == 9u);
+
+    /* The "!" stop token terminates the lexer. The context input text position
+     * must not be updated when the lexer user code requests termination, so it
+     * still points at the "!" token on row 3, col 1. */
+    assert(p_lex(context, &token_info) == P_USER_TERMINATED);
+    assert(p_user_terminate_code(context) == 42u);
+    assert(context->text_position.row == 3u);
+    assert(context->text_position.col == 1u);
+
+    p_context_delete(context);
+    return 0;
+}

spec/test_lexer_positions.d

@@ -0,0 +1,42 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "abc\n  defg hi\n!";
+    p_context_t * context = p_context_new(input);
+    p_token_info_t token_info;
+
+    /* First token "abc" on row 1, cols 1-3. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context.last_start == p_position_t(1, 1));
+    assert(context.last_end == p_position_t(1, 3));
+    /* The lexer code block observed the same positions reported to the caller. */
+    assert(context.last_start == token_info.position);
+    assert(context.last_end == token_info.end_position);
+
+    /* Second token "defg" on row 2, cols 3-6. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context.last_start == p_position_t(2, 3));
+    assert(context.last_end == p_position_t(2, 6));
+
+    /* Third token "hi" on row 2, cols 8-9. */
+    assert(p_lex(context, &token_info) == P_SUCCESS);
+    assert(token_info.token == TOKEN_word);
+    assert(context.last_start == p_position_t(2, 8));
+    assert(context.last_end == p_position_t(2, 9));
+
+    /* The "!" stop token terminates the lexer. The context input text position
+     * must not be updated when the lexer user code requests termination, so it
+     * still points at the "!" token on row 3, col 1. */
+    assert(p_lex(context, &token_info) == P_USER_TERMINATED);
+    assert(p_user_terminate_code(context) == 42u);
+    assert(context.text_position == p_position_t(3, 1));
+}

spec/test_lexer_result_value.c

@@ -5,15 +5,17 @@
 int main()
 {
     char const * input = "x";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 1u);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 1u);
+    p_context_delete(context);
 
     input = "fabulous";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 8u);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 8u);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer_result_value.d

@@ -9,13 +9,13 @@ int main()
 unittest
 {
     string input = `x`;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 1u);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 1u);
 
     input = `fabulous`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 8u);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 8u);
 }

spec/test_lexer_unknown_character.c

@@ -5,14 +5,16 @@
 int main()
 {
     char const * input = "x";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_UNEXPECTED_INPUT);
+    p_context_delete(context);
 
     input = "123";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 123u);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 123u);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_lexer_unknown_character.d

@@ -9,12 +9,12 @@ int main()
 unittest
 {
     string input = `x`;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_UNEXPECTED_INPUT);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_UNEXPECTED_INPUT);
 
     input = `123`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 123u);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 123u);
 }

spec/test_match_backslashes.c

@@ -5,9 +5,10 @@
 int main()
 {
     char const * input = "\a\b\t\n\v\f\rt";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_match_backslashes.d

@@ -9,7 +9,7 @@ int main()
 unittest
 {
     string input = "\a\b\t\n\v\f\rt";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_multiple_parsers.c

@@ -6,14 +6,16 @@
 int main()
 {
     char const * input1 = "a\n1";
-    myp1_context_t context1;
-    myp1_context_init(&context1, (uint8_t const *)input1, strlen(input1));
-    assert(myp1_parse(&context1) == MYP1_SUCCESS);
+    myp1_context_t * context1;
+    context1 = myp1_context_new((uint8_t const *)input1, strlen(input1));
+    assert(myp1_parse(context1) == MYP1_SUCCESS);
+    myp1_context_delete(context1);
 
     char const * input2 = "bcb";
-    myp2_context_t context2;
-    myp2_context_init(&context2, (uint8_t const *)input2, strlen(input2));
-    assert(myp2_parse(&context2) == MYP2_SUCCESS);
+    myp2_context_t * context2;
+    context2 = myp2_context_new((uint8_t const *)input2, strlen(input2));
+    assert(myp2_parse(context2) == MYP2_SUCCESS);
+    myp2_context_delete(context2);
 
     return 0;
 }

spec/test_multiple_parsers.d

@@ -10,12 +10,12 @@ int main()
 unittest
 {
     string input1 = "a\n1";
-    myp1_context_t context1;
-    myp1_context_init(&context1, input1);
-    assert(myp1_parse(&context1) == MYP1_SUCCESS);
+    myp1_context_t * context1;
+    context1 = myp1_context_new(input1);
+    assert(myp1_parse(context1) == MYP1_SUCCESS);
 
     string input2 = "bcb";
-    myp2_context_t context2;
-    myp2_context_init(&context2, input2);
-    assert(myp2_parse(&context2) == MYP2_SUCCESS);
+    myp2_context_t * context2;
+    context2 = myp2_context_new(input2);
+    assert(myp2_parse(context2) == MYP2_SUCCESS);
 }

spec/test_named_optional_rule_component_tree.c

@@ -6,10 +6,10 @@
 int main()
 {
     char const * input = "b";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
     assert(start->a == NULL);
     assert(start->pToken2 != NULL);
     assert_eq(TOKEN_b, start->pToken2->token);
@@ -17,10 +17,13 @@ int main()
     assert(start->pR == NULL);
     assert(start->r == NULL);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "abcd";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start->a != NULL);
     assert_eq(TOKEN_a, start->pToken1->token);
     assert(start->pToken2 != NULL);
@@ -31,15 +34,21 @@ int main()
     assert(start->pR == start->r);
     assert_eq(TOKEN_c, start->pR->pToken1->token);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "bdc";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start->a == NULL);
     assert(start->pToken2 != NULL);
     assert(start->r != NULL);
     assert_eq(TOKEN_d, start->pR->pToken1->token);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     return 0;
 }
 

spec/test_named_optional_rule_component_tree.d

@@ -10,10 +10,9 @@ int main()
 unittest
 {
     string input = "b";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
     assert(start.pToken1 is null);
     assert(start.pToken2 !is null);
     assert_eq(TOKEN_b, start.pToken2.token);
@@ -21,10 +20,12 @@ unittest
     assert(start.pR is null);
     assert(start.r is null);
 
+    p_tree_delete(start);
+
     input = "abcd";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start.pToken1 != null);
     assert_eq(TOKEN_a, start.pToken1.token);
     assert(start.pToken2 != null);
@@ -35,12 +36,16 @@ unittest
     assert(start.pR == start.r);
     assert_eq(TOKEN_c, start.pR.pToken1.token);
 
+    p_tree_delete(start);
+
     input = "bdc";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start.pToken1 is null);
     assert(start.pToken2 !is null);
     assert(start.pR !is null);
     assert_eq(TOKEN_d, start.pR.pToken1.token);
+
+    p_tree_delete(start);
 }

spec/test_optional_rule_component.c

@@ -5,17 +5,20 @@
 int main()
 {
     char const * input = "b";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "abcd";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "abdc";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_optional_rule_component.d

@@ -9,15 +9,15 @@ int main()
 unittest
 {
     string input = "b";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = "abcd";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = "abdc";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_optional_rule_component_tree.c

@@ -6,20 +6,23 @@
 int main()
 {
     char const * input = "b";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
     assert(start->pToken1 == NULL);
     assert(start->pToken2 != NULL);
     assert_eq(TOKEN_b, start->pToken2->token);
     assert(start->pR3 == NULL);
     assert(start->pR == NULL);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "abcd";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start->pToken1 != NULL);
     assert_eq(TOKEN_a, start->pToken1->token);
     assert(start->pToken2 != NULL);
@@ -28,15 +31,21 @@ int main()
     assert(start->pR == start->pR3);
     assert_eq(TOKEN_c, start->pR->pToken1->token);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "bdc";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start->pToken1 == NULL);
     assert(start->pToken2 != NULL);
     assert(start->pR != NULL);
     assert_eq(TOKEN_d, start->pR->pToken1->token);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     return 0;
 }
 

spec/test_optional_rule_component_tree.d

@@ -10,20 +10,21 @@ int main()
 unittest
 {
     string input = "b";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
     assert(start.pToken1 is null);
     assert(start.pToken2 !is null);
     assert_eq(TOKEN_b, start.pToken2.token);
     assert(start.pR3 is null);
     assert(start.pR is null);
 
+    p_tree_delete(start);
+
     input = "abcd";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start.pToken1 != null);
     assert_eq(TOKEN_a, start.pToken1.token);
     assert(start.pToken2 != null);
@@ -32,12 +33,16 @@ unittest
     assert(start.pR == start.pR3);
     assert_eq(TOKEN_c, start.pR.pToken1.token);
 
+    p_tree_delete(start);
+
     input = "bdc";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    start = p_result(context);
     assert(start.pToken1 is null);
     assert(start.pToken2 !is null);
     assert(start.pR !is null);
     assert_eq(TOKEN_d, start.pR.pToken1.token);
+
+    p_tree_delete(start);
 }

spec/test_parser_identical_rules_lookahead.c

@@ -5,13 +5,15 @@
 int main()
 {
     char const * input = "aba";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "abb";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_parser_identical_rules_lookahead.d

@@ -9,11 +9,11 @@ int main()
 unittest
 {
     string input = "aba";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = "abb";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_parser_rule_from_multiple_states.c

@@ -5,20 +5,23 @@
 int main()
 {
     char const * input = "a";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context).row == 1);
-    assert(p_position(&context).col == 2);
-    assert(context.token == TOKEN___EOF);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context).row == 1);
+    assert(p_position(context).col == 2);
+    assert(context->token == TOKEN___EOF);
+    p_context_delete(context);
 
     input = "a b";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "bb";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_parser_rule_from_multiple_states.d

@@ -9,17 +9,17 @@ int main()
 unittest
 {
     string input = "a";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_UNEXPECTED_TOKEN);
-    assert(p_position(&context) == p_position_t(1, 2));
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_UNEXPECTED_TOKEN);
+    assert(p_position(context) == p_position_t(1, 2));
     assert(context.token == TOKEN___EOF);
 
     input = "a b";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = "bb";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_parser_rule_user_code.c

@@ -5,9 +5,10 @@
 int main()
 {
     char const * input = "ab";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_parser_rule_user_code.d

@@ -9,7 +9,7 @@ int main()
 unittest
 {
     string input = "ab";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_parser_user_code_tree.c

@@ -0,0 +1,39 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include "testutils.h"
+
+int main()
+{
+    char const * input = "ab";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+
+    /* The parser user code recorded values accessed via $$, $1, and $2 while
+     * the tree node for the Start rule was being formed. */
+    assert_eq(3, context->start_n_fields);
+    assert_eq(11, context->start_a_value);
+    assert_eq(11, context->a_value);
+    assert_eq(22, context->b_value);
+    assert_eq(TOKEN_b, context->b_token);
+
+    /* The empty-matched rule C has a null $$ tree node, and its field in the
+     * Start node is null as well. */
+    assert_eq(1, context->c_is_null);
+    assert_eq(1, context->c_field_is_null);
+
+    /* Field aliases reference the same component tree nodes as the positional
+     * references. */
+    assert_eq(11, context->alias_a_value);
+    assert_eq(22, context->alias_b_value);
+
+    Start * start = p_result(context);
+    assert(start->pA != NULL);
+    assert(start->pB != NULL);
+    assert(start->pC == NULL);
+    p_tree_delete(start);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_parser_user_code_tree.d

@@ -0,0 +1,39 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "ab";
+    p_context_t * context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+
+    /* The parser user code recorded values accessed via $$, $1, and $2 while
+     * the tree node for the Start rule was being formed. */
+    assert_eq(3, context.start_n_fields);
+    assert_eq(11, context.start_a_value);
+    assert_eq(11, context.a_value);
+    assert_eq(22, context.b_value);
+    assert_eq(TOKEN_b, context.b_token);
+
+    /* The empty-matched rule C has a null $$ tree node, and its field in the
+     * Start node is null as well. */
+    assert_eq(1, context.c_is_null);
+    assert_eq(1, context.c_field_is_null);
+
+    /* Field aliases reference the same component tree nodes as the positional
+     * references. */
+    assert_eq(11, context.alias_a_value);
+    assert_eq(22, context.alias_b_value);
+
+    Start * start = p_result(context);
+    assert(start.pA !is null);
+    assert(start.pB !is null);
+    assert(start.pC is null);
+    p_tree_delete(start);
+}

spec/test_parsing_json.c

@@ -6,51 +6,58 @@
 int main()
 {
     char const * input = "";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     input = "{}";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context)->id == JSON_OBJECT);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context)->id == JSON_OBJECT);
+    p_context_delete(context);
 
     input = "[]";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context)->id == JSON_ARRAY);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context)->id == JSON_ARRAY);
+    p_context_delete(context);
 
     input = "-45.6";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context)->id == JSON_NUMBER);
-    assert(p_result(&context)->number == -45.6);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context)->id == JSON_NUMBER);
+    assert(p_result(context)->number == -45.6);
+    p_context_delete(context);
 
     input = "2E-2";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context)->id == JSON_NUMBER);
-    assert(p_result(&context)->number == 0.02);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context)->id == JSON_NUMBER);
+    assert(p_result(context)->number == 0.02);
+    p_context_delete(context);
 
     input = "{\"hi\":true}";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    JSONValue * o = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    JSONValue * o = p_result(context);
     assert(o->id == JSON_OBJECT);
     assert_eq(1, o->object.size);
     assert(strcmp(o->object.entries[0].name, "hi") == 0);
     assert(o->object.entries[0].value->id == JSON_TRUE);
+    p_context_delete(context);
 
     input = "{\"ff\": false, \"nn\": null}";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    o = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    o = p_result(context);
     assert(o->id == JSON_OBJECT);
     assert_eq(2, o->object.size);
     assert(strcmp(o->object.entries[0].name, "ff") == 0);
     assert(o->object.entries[0].value->id == JSON_FALSE);
     assert(strcmp(o->object.entries[1].name, "nn") == 0);
     assert(o->object.entries[1].value->id == JSON_NULL);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_parsing_json.d

@@ -10,45 +10,45 @@ int main()
 unittest
 {
     string input = ``;
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 
     input = `{}`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONObject)p_result(&context));
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(context));
 
     input = `[]`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONArray)p_result(&context));
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONArray)p_result(context));
 
     input = `-45.6`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONNumber)p_result(&context));
-    assert((cast(JSONNumber)p_result(&context)).value == -45.6);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONNumber)p_result(context));
+    assert((cast(JSONNumber)p_result(context)).value == -45.6);
 
     input = `2E-2`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONNumber)p_result(&context));
-    assert((cast(JSONNumber)p_result(&context)).value == 0.02);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONNumber)p_result(context));
+    assert((cast(JSONNumber)p_result(context)).value == 0.02);
 
     input = `{"hi":true}`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONObject)p_result(&context));
-    JSONObject o = cast(JSONObject)p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(context));
+    JSONObject o = cast(JSONObject)p_result(context);
     assert(o.value["hi"]);
     assert(cast(JSONTrue)o.value["hi"]);
 
     input = `{"ff": false, "nn": null}`;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(cast(JSONObject)p_result(&context));
-    o = cast(JSONObject)p_result(&context);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(cast(JSONObject)p_result(context));
+    o = cast(JSONObject)p_result(context);
     assert(o.value["ff"]);
     assert(cast(JSONFalse)o.value["ff"]);
     assert(o.value["nn"]);

spec/test_parsing_lists.c

@@ -5,20 +5,23 @@
 int main()
 {
     char const * input = "a";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 1u);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 1u);
+    p_context_delete(context);
 
     input = "";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 0u);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 0u);
+    p_context_delete(context);
 
     input = "aaaaaaaaaaaaaaaa";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 16u);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 16u);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_parsing_lists.d

@@ -9,18 +9,18 @@ int main()
 unittest
 {
     string input = "a";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 1u);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 1u);
 
     input = "";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 0u);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 0u);
 
     input = "aaaaaaaaaaaaaaaa";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    assert(p_result(&context) == 16u);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    assert(p_result(context) == 16u);
 }

spec/test_pattern.c

@@ -6,15 +6,17 @@
 int main()
 {
     char const * input = "abcdef";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass1\n");
+    p_context_delete(context);
 
     input = "defabcdef";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
     printf("pass2\n");
+    p_context_delete(context);
 
     return 0;
 }

spec/test_pattern.d

@@ -9,13 +9,13 @@ int main()
 unittest
 {
     string input = "abcdef";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass1");
 
     input = "defabcdef";
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
     writeln("pass2");
 }

spec/test_positions.c

@@ -0,0 +1,29 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include <stdio.h>
+
+int main()
+{
+    char const * input = "    Hello\n\n        4200\n";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
+
+    printf("\n");
+
+    input = "\n tok2";
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
+
+    printf("\n");
+
+    input = "  tok1";
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_positions.d

@@ -0,0 +1,27 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "    Hello\n\n        4200\n";
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+
+    writeln();
+
+    input = "\n tok2";
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+
+    writeln();
+
+    input = "  tok1";
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+}

spec/test_return_token_from_pattern.c

@@ -5,9 +5,10 @@
 int main()
 {
     char const * input = "defghidef";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    p_context_delete(context);
 
     return 0;
 }

spec/test_return_token_from_pattern.d

@@ -9,7 +9,7 @@ int main()
 unittest
 {
     string input = "defghidef";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
 }

spec/test_start_rule_ast.c

@@ -1,17 +0,0 @@
-#include "testparser.h"
-#include <assert.h>
-#include <string.h>
-#include "testutils.h"
-
-int main()
-{
-    char const * input = "hi";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    Top * top = p_result(&context);
-    assert(top->pToken != NULL);
-    assert_eq(TOKEN_hi, top->pToken->token);
-
-    return 0;
-}

spec/test_start_rule_tree.c

@@ -0,0 +1,20 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include "testutils.h"
+
+int main()
+{
+    char const * input = "hi";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    Top * top = p_result(context);
+    assert(top->pToken != NULL);
+    assert_eq(TOKEN_hi, top->pToken->token);
+
+    p_tree_delete(top);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_start_rule_tree.d

@@ -10,10 +10,10 @@ int main()
 unittest
 {
     string input = "hi";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    Top * top = p_result(&context);
+    p_context_t * context;
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    Top * top = p_result(context);
     assert(top.pToken !is null);
     assert_eq(TOKEN_hi, top.pToken.token);
 }

spec/test_starting_rules.c

@@ -0,0 +1,30 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include "testutils.h"
+
+int main()
+{
+    char const * input = "bbbb";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    int result = p_result(context);
+    assert_eq(8, result);
+    p_context_delete(context);
+
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse_Bs(context) == P_SUCCESS);
+    result = p_result_Bs(context);
+    assert_eq(8, result);
+    p_context_delete(context);
+
+    input = "c";
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse_R(context) == P_SUCCESS);
+    result = p_result_R(context);
+    assert_eq(3, result);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_starting_rules.d

@@ -0,0 +1,29 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "bbbb";
+    p_context_t * context;
+    context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    int result = p_result(context);
+    assert(result == 8);
+
+    context = p_context_new(input);
+    assert(p_parse_Bs(context) == P_SUCCESS);
+    result = p_result_Bs(context);
+    assert(result == 8);
+
+    input = "c";
+    context = p_context_new(input);
+    assert(p_parse_R(context) == P_SUCCESS);
+    result = p_result_R(context);
+    assert(result == 3);
+}

spec/test_starting_rules_tree.c

@@ -0,0 +1,40 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include "testutils.h"
+
+int main()
+{
+    char const * input = "bbbb";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
+    assert_not_null(start->bs);
+    assert_not_null(start->bs->b);
+    assert_not_null(start->bs->bs->b);
+    assert_not_null(start->bs->bs->bs->b);
+    assert_not_null(start->bs->bs->bs->bs->b);
+    p_tree_delete(start);
+    p_context_delete(context);
+
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse_Bs(context) == P_SUCCESS);
+    Bs * bs = p_result_Bs(context);
+    assert_not_null(bs->b);
+    assert_not_null(bs->bs->b);
+    assert_not_null(bs->bs->bs->b);
+    assert_not_null(bs->bs->bs->bs->b);
+    p_tree_delete_Bs(bs);
+    p_context_delete(context);
+
+    input = "c";
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse_R(context) == P_SUCCESS);
+    R * r = p_result_R(context);
+    assert_not_null(r->c);
+    p_tree_delete_R(r);
+    p_context_delete(context);
+
+    return 0;
+}

spec/test_starting_rules_tree.d

@@ -0,0 +1,41 @@
+import testparser;
+import std.stdio;
+import testutils;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input = "bbbb";
+    p_context_t * context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
+    assert(start.bs);
+    assert(start.bs.b);
+    assert(start.bs.bs.b);
+    assert(start.bs.bs.bs.b);
+    assert(start.bs.bs.bs.bs.b);
+
+    p_tree_delete(start);
+
+    context = p_context_new(input);
+    assert(p_parse_Bs(context) == P_SUCCESS);
+    Bs * bs = p_result_Bs(context);
+    assert(bs.b);
+    assert(bs.bs.b);
+    assert(bs.bs.bs.b);
+    assert(bs.bs.bs.bs.b);
+
+    p_tree_delete_Bs(bs);
+
+    input = "c";
+    context = p_context_new(input);
+    assert(p_parse_R(context) == P_SUCCESS);
+    R * r = p_result_R(context);
+    assert(r.c);
+
+    p_tree_delete_R(r);
+}

spec/test_token_user_fields.c

@@ -0,0 +1,46 @@
+#include "testparser.h"
+#include <assert.h>
+#include <stdio.h>
+#include <string.h>
+#include <stdlib.h>
+
+int main()
+{
+    char const * input =
+        "# c1\n"
+        "#  c2\n"
+        "\n"
+        "first\n"
+        "\n   \n  \n"
+        "  # s1\n"
+        "   #   s2\n"
+        "second\n";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
+    assert(start->pIDs);
+    assert(start->pIDs->id);
+#ifdef __cplusplus
+    assert(start->pIDs->id->comments == "# c1\n#  c2\n");
+#else
+    assert(start->pIDs->id->comments);
+    assert(strcmp(start->pIDs->id->comments, "# c1\n#  c2\n") == 0);
+#endif
+    assert(start->pIDs->pIDs);
+    assert(start->pIDs->pIDs->id);
+#ifdef __cplusplus
+    assert(start->pIDs->pIDs->id->comments == "# s1\n#   s2\n");
+#else
+    assert(start->pIDs->pIDs->id->comments);
+    assert(strcmp(start->pIDs->pIDs->id->comments, "# s1\n#   s2\n") == 0);
+#endif
+
+#ifndef __cplusplus
+    free(context->comments);
+#endif
+    p_context_delete(context);
+    p_tree_delete(start);
+
+    return 0;
+}

spec/test_token_user_fields.d

@@ -0,0 +1,31 @@
+import testparser;
+import std.stdio;
+
+int main()
+{
+    return 0;
+}
+
+unittest
+{
+    string input =
+        "# c1\n" ~
+        "#  c2\n" ~
+        "\n" ~
+        "first\n" ~
+        "\n   \n  \n" ~
+        "  # s1\n" ~
+        "   #   s2\n" ~
+        "second\n";
+    p_context_t * context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
+    assert(start.pIDs);
+    assert(start.pIDs.id);
+    assert(start.pIDs.id.comments == "# c1\n#  c2\n");
+    assert(start.pIDs.pIDs);
+    assert(start.pIDs.pIDs.id);
+    assert(start.pIDs.pIDs.id.comments == "# s1\n#   s2\n");
+
+    p_tree_delete(start);
+}

spec/test_tree.c

@@ -6,10 +6,10 @@
 int main()
 {
     char const * input = "a, ((b)), b";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    Start * start = p_result(&context);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    Start * start = p_result(context);
     assert(start->pItems1 != NULL);
     assert(start->pItems != NULL);
     Items * items = start->pItems;
@@ -33,16 +33,22 @@ int main()
     assert_eq(22, itemsmore->pItem->pToken1->pvalue);
     assert(itemsmore->pItemsMore == NULL);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    start = p_result(context);
     assert(start->pItems == NULL);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     input = "2 1";
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert_eq(P_SUCCESS, p_parse(&context));
-    start = p_result(&context);
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    start = p_result(context);
     assert(start->pItems != NULL);
     assert(start->pItems->pItem != NULL);
     assert(start->pItems->pItem->pDual != NULL);
@@ -51,5 +57,8 @@ int main()
     assert(start->pItems->pItem->pDual->pTwo2 == NULL);
     assert(start->pItems->pItem->pDual->pOne1 == NULL);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     return 0;
 }

spec/test_tree.d

@@ -10,10 +10,9 @@ int main()
 unittest
 {
     string input = "a, ((b)), b";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    Start * start = p_result(&context);
+    p_context_t * context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    Start * start = p_result(context);
     assert(start.pItems1 !is null);
     assert(start.pItems !is null);
     Items * items = start.pItems;
@@ -37,16 +36,20 @@ unittest
     assert_eq(22, itemsmore.pItem.pToken1.pvalue);
     assert(itemsmore.pItemsMore is null);
 
+    p_tree_delete(start);
+
     input = "";
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    start = p_result(context);
     assert(start.pItems is null);
 
+    p_tree_delete(start);
+
     input = "2 1";
-    p_context_init(&context, input);
-    assert_eq(P_SUCCESS, p_parse(&context));
-    start = p_result(&context);
+    context = p_context_new(input);
+    assert_eq(P_SUCCESS, p_parse(context));
+    start = p_result(context);
     assert(start.pItems !is null);
     assert(start.pItems.pItem !is null);
     assert(start.pItems.pItem.pDual !is null);
@@ -54,4 +57,6 @@ unittest
     assert(start.pItems.pItem.pDual.pOne2 !is null);
     assert(start.pItems.pItem.pDual.pTwo2 is null);
     assert(start.pItems.pItem.pDual.pOne1 is null);
+
+    p_tree_delete(start);
 }

spec/test_tree_delete_token_node_memory.c

@@ -0,0 +1,20 @@
+#include "testparser.h"
+#include <assert.h>
+#include <string.h>
+#include "testutils.h"
+
+int main()
+{
+    char const * input = "ab";
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert_eq(P_SUCCESS, p_parse(context));
+    Start * start = p_result(context);
+    assert(start->a != NULL);
+    assert(*start->a->pvalue == 1);
+    assert(start->b != NULL);
+    assert(*start->b->pvalue == 2);
+
+    p_tree_delete(start);
+    p_context_delete(context);
+}

spec/test_tree_field_aliases.c

@@ -6,14 +6,17 @@
 int main()
 {
     char const * input = "\na\nb\nc";
-    p_context_t context;
-    p_context_init(&context, (uint8_t const *)input, strlen(input));
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context;
+    context = p_context_new((uint8_t const *)input, strlen(input));
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
 
     assert_eq(TOKEN_a, start->first->pToken->token);
     assert_eq(TOKEN_b, start->second->pToken->token);
     assert_eq(TOKEN_c, start->third->pToken->token);
 
+    p_tree_delete(start);
+    p_context_delete(context);
+
     return 0;
 }

spec/test_tree_field_aliases.d

@@ -10,12 +10,13 @@ int main()
 unittest
 {
     string input = "\na\nb\nc";
-    p_context_t context;
-    p_context_init(&context, input);
-    assert(p_parse(&context) == P_SUCCESS);
-    Start * start = p_result(&context);
+    p_context_t * context = p_context_new(input);
+    assert(p_parse(context) == P_SUCCESS);
+    Start * start = p_result(context);
 
     assert_eq(TOKEN_a, start.first.pToken.token);
     assert_eq(TOKEN_b, start.second.pToken.token);
     assert_eq(TOKEN_c, start.third.pToken.token);
+
+    p_tree_delete(start);
 }